Fundamentals of Robotics Level 2 or equivalent β including basic programming, sensor/actuator integration, motion systems, and autonomous navigation.
Duration: 12β16 weeks (one semester) | Target Audience: Students aged 15+ who have completed Level 2 and are ready for advanced robotics concepts.
Level 3 transitions from foundational robotics to advanced autonomy, learning, and interaction. You will move beyond simple obstacle avoidance into probabilistic robotics, robot learning, manipulation, and human-robot interaction. The course emphasizes both theoretical foundations (kinematics, control, probability) and hands-on implementation using ROS and simulation platforms.
By the end of this course, students will be able to:
| Module | Title | Duration | Focus |
|---|---|---|---|
| 1 | Advanced Kinematics and Dynamics | 2 weeks | Configuration space, singularities, dynamics |
| 2 | Probabilistic Robotics | 2 weeks | Bayesian filtering, localization, SLAM |
| 3 | Motion Planning | 2 weeks | Sampling-based planning, trajectory optimization |
| 4 | Robot Learning | 2 weeks | Reinforcement learning, imitation learning |
| 5 | Manipulation and Grasping | 2 weeks | Contact modeling, dexterous manipulation |
| 6 | Human-Robot Interaction | 2 weeks | Shared control, safety, cognitive robotics |
| 7 | Advanced Application Domains | 2 weeks | Aerial, underwater, soft robotics |
| 8 | Capstone Project | 2β4 weeks | Full system integration |
| Component | Weight | Description |
|---|---|---|
| Module Assignments | 40% | 6β8 hands-on programming assignments |
| Midterm Project | 15% | Integrated system demonstrating core concepts |
| Capstone Project | 30% | Full autonomous system with documentation |
| Participation | 15% | Lab engagement, peer collaboration, discussion |
| Category | Tools |
|---|---|
| Programming | Python, C++, MATLAB/Octave |
| Robotics Framework | ROS (Robot Operating System) |
| Simulation | Gazebo, PyBullet, Webots |
| CAD | SolidWorks, Fusion 360 (for mechanical design) |
| Hardware | Arduino, Raspberry Pi, NVIDIA Jetson, mobile manipulation platforms |
Before starting Level 3, ensure you are comfortable with:
| Track | Next Steps |
|---|---|
| Research | Graduate studies in robotics, AI, or control systems |
| Industry | Autonomous systems, warehouse robotics, drone development |
| Entrepreneurship | Robotics startups, automation solutions |
| Specialization | Medical robotics, space robotics, agricultural automation |
Welcome to Level Three of Fundamentals of Robotics! You have come a very long way. In Level Two, you learned how to program robots, use sensors, control motors, navigate autonomously, integrate systems, design robot bodies, and complete a capstone project. Now you are ready for something more advanced.
In this module, you will learn about kinematics and dynamics. These are big words, but do not worry. We will explain them in simple language.
Kinematics is the study of motion without worrying about forces. It answers questions like: Where is the robot's hand? How fast is it moving? What angle is the joint at?
Dynamics is the study of motion with forces. It answers questions like: How much force is needed to lift this object? What happens when the robot pushes something? Why does the robot slow down when going uphill?
These ideas are the foundation of advanced robotics. They are used in factory robots, surgical robots, self-driving cars, and space robots. If you want to build real, professional robots, you need to understand kinematics and dynamics.
We will use lots of examples, stories, and illustrations to make these ideas clear. By the end of this module, you will understand how robots move and why they move the way they do.
Let us begin!
By the end of this module, you will be able to:
Once upon a time, in the city of Lagos, Nigeria, there lived a girl named Ada. Ada had completed Fundamentals of Robotics Level Two. She could program robots, use sensors, and even build her own robot chassis. She was ready for something bigger.
One day, Ada's uncle invited her to visit a factory in Ikeja. The factory made electronics. It had many robot arms that picked up tiny parts and placed them on circuit boards.
Ada watched the robot arm in amazement. It moved smoothly. It reached for a tiny chip. It picked it up. It placed it exactly on the board. It did this over and over, hundreds of times, without making a mistake.
"Uncle, how does the robot know where to move?" Ada asked.
"That is called kinematics," her uncle said. "Kinematics is the study of motion. The robot's computer calculates where each joint should be so that the hand reaches the right spot."
"But how does it know how much force to use?" Ada asked. "The chip is very small. It could crush it."
"That is called dynamics," her uncle said. "Dynamics is the study of forces. The robot's computer calculates how much force is needed to pick up the chip without breaking it."
Ada was fascinated. She asked more questions. How does the robot know its own position? How does it avoid hitting itself? What happens if two joints try to move to the same place?
Her uncle smiled. "Ada, you are asking questions about advanced kinematics. You should study Level Three."
Ada went home and started learning. She learned about configuration space. She learned about degrees of freedom. She learned about forward kinematics and inverse kinematics. She learned about Jacobian matrices and singularities.
At first, the ideas were hard. But Ada did not give up. She drew diagrams. She worked through examples. She asked questions. Slowly, things became clear.
Months later, Ada built a small robot arm. She used her knowledge of kinematics to calculate joint angles. She used her knowledge of dynamics to control the force. The arm could pick up a small ball and place it in a cup. It was not as fast as the factory robot, but it worked.
Ada smiled. She had taken her first step into advanced robotics.
That is what you will do in this module. You will learn about advanced kinematics and dynamics. You will take your first step into professional robotics.
Let us begin!
Kinematics is the study of motion. It looks at position, velocity, and acceleration. It does not worry about forces.
Kinematics tells us where a robot is and how it moves. Without kinematics, we cannot plan robot movements. We cannot tell a robot arm to reach for an object. We cannot make a robot drive to a destination.
Imagine you are playing football. You kick the ball. Kinematics asks: Where did the ball go? How fast was it going? What path did it follow? It does not ask why the ball moved. It just describes the motion.
When you throw a ball, kinematics describes its path. It tells you how high it goes and where it lands.
In physics class, you learn about motion. How fast does a car go? How far does it travel? That is kinematics.
When you walk from your room to the kitchen, kinematics describes your path. How far did you walk? How long did it take?
A danfo bus travels from Yaba to Ikeja. Kinematics describes the distance and time.
Kinematics Start | | (path) V End Kinematics describes: - Position - Velocity - Acceleration
Kinematics is the study of motion. It looks at position, velocity, and acceleration. It does not worry about forces.
Dynamics is the study of motion with forces. It looks at how forces affect movement.
Dynamics tells us how much force is needed to move a robot. It tells us how heavy objects affect the robot. Without dynamics, the robot might use too much force or too little force. It might break things or fail to lift them.
Imagine pushing a box. If the box is light, you push gently. If the box is heavy, you push harder. Dynamics is about how much force is needed.
A car needs more force to go uphill than downhill. Dynamics explains why.
In physics class, you learn about forces. Gravity pulls things down. Friction slows things down. That is dynamics.
When you push a door, you use force. Dynamics describes how much force is needed.
A trader pushes a wheelbarrow full of goods. Dynamics describes how much force is needed.
Dynamics Force --> Object --> Motion Dynamics looks at: - Force - Mass - Acceleration - Gravity - Friction
Dynamics is the study of motion with forces. It looks at how forces affect movement.
Configuration space is a way to describe all possible positions of a robot. Each point in configuration space represents one position of the robot.
Configuration space helps us plan robot movements. We can see which positions are free and which are blocked. We can find a path from one position to another.
Imagine a robot arm with one joint. The joint can rotate from 0 degrees to 180 degrees. The configuration space is all the angles between 0 and 180. Each angle is one configuration.
When you open a door, the door can be at many angles. Each angle is a configuration.
A pencil can point in many directions. Each direction is a configuration.
A fan can rotate to many angles. Each angle is a configuration.
A traffic warden can point in many directions. Each direction is a configuration.
Configuration Space
For a 1-joint robot:
0 degrees
|
V
45 degrees
|
V
90 degrees
|
V
135 degrees
|
V
180 degrees
Each angle is a configuration.
Configuration space describes all possible positions of a robot. Each point is one configuration.
Degrees of freedom (DOF) is the number of independent ways a robot can move. Each joint adds one or more degrees of freedom.
Degrees of freedom tell us how flexible a robot is. A robot with more DOF can reach more places. A robot with fewer DOF is simpler but less flexible.
Think of your arm. You can move it up and down. You can move it left and right. You can rotate it. Each of these movements is a degree of freedom.
| Robot | Degrees of Freedom | Movement |
|---|---|---|
| Simple wheeled robot | 2 | Forward/backward, turn |
| Robot arm (3 joints) | 3 | Reach in 3D space |
| Human arm | 7 | Very flexible |
| Drone | 6 | Move in all directions |
Your shoulder has 3 degrees of freedom. Your elbow has 1. Your wrist has 2.
A pair of scissors has 1 degree of freedom. It only opens and closes.
A door has 1 degree of freedom. It only swings open and closed.
A keke napep has 2 degrees of freedom. It moves forward and turns.
Degrees of Freedom 1 DOF: +---+ | | <-- opens and closes +---+ 2 DOF: +---+ | | <-- opens/closes and slides +---+ 3 DOF: +---+ | | <-- opens/closes, slides, and rotates +---+
Degrees of freedom is the number of independent ways a robot can move. More DOF means more flexibility.
Forward kinematics is finding the position of the robot's hand (or end effector) when we know the joint angles.
Forward kinematics tells us where the robot's hand is. It helps us check if the robot will hit something. It helps us plan movements.
Imagine your arm. If you know the angle of your shoulder and the angle of your elbow, you can calculate where your hand is. That is forward kinematics.
A crane operator knows the angle of the crane arm. Forward kinematics tells him where the hook is.
If you know the length of a pencil and the angle you hold it, you can calculate where the tip is.
If you know the angle of a door and the width of the door, you can calculate where the edge is.
If you know the angle of a borehole pipe, you can calculate where the water will come out.
Forward Kinematics
Joint 1 angle: 30 degrees
Joint 2 angle: 45 degrees
Joint 1
/|
/ |
/ |
/ |
/ |
/ |
/ |
Joint 2 |
\ |
\ |
\ |
\ |
\ |
\ |
\|
End Effector
Calculate position of end effector.
Forward kinematics finds the position of the robot's hand when we know the joint angles.
Inverse kinematics is finding the joint angles when we know the position of the robot's hand. It is the opposite of forward kinematics.
Inverse kinematics is used to control robots. We tell the robot where to put its hand. The robot calculates the joint angles needed.
Imagine you want to touch a spot on the wall. Your brain calculates the angles of your shoulder and elbow. That is inverse kinematics.
A robot arm picks up a bottle. The robot knows where the bottle is. Inverse kinematics calculates the joint angles.
You want to write on the board. Your brain calculates the angles of your arm.
You want to reach for a cup. Your brain calculates the angles of your arm.
A farmer wants to reach a fruit on a tree. His brain calculates the angles of his arm.
Inverse Kinematics
Desired position: (10, 20)
|
V
Calculate joint angles
|
V
Joint 1: 35 degrees
Joint 2: 50 degrees
|
V
Move robot arm
Inverse kinematics finds the joint angles when we know the position of the robot's hand.
The Jacobian matrix is a mathematical tool that relates joint velocities to end-effector velocities. It tells us how fast the hand moves when the joints move.
The Jacobian is used to control robots precisely. It helps us calculate how much to move each joint to get the hand to move in a certain direction.
Imagine a bicycle. If you pedal faster, the wheels turn faster. The Jacobian tells you how pedaling speed affects wheel speed.
A crane operator knows that moving the crane arm slightly moves the hook a lot. The Jacobian describes this relationship.
A lever helps you lift heavy things. The Jacobian describes how lever movement affects the load.
A can opener has a handle. The Jacobian describes how handle movement affects the blade.
A mortar and pestle. The Jacobian describes how pestle movement affects the grinding.
Jacobian Matrix
Joint velocities:
[dΞΈ1/dt]
[dΞΈ2/dt]
|
| (Jacobian)
V
End-effector velocities:
[dx/dt]
[dy/dt]
The Jacobian matrix relates joint velocities to end-effector velocities. It is used for precise robot control.
A singularity is a configuration where the robot loses some ability to move. At a singularity, the Jacobian matrix becomes singular (it cannot be inverted).
Singularities cause problems. The robot might not be able to move in a certain direction. It might need infinite joint velocities. It might become unstable.
Imagine your arm is fully stretched out. You cannot stretch it further. That is a singularity. You have lost the ability to reach further.
A door that is fully open cannot open more. That is a singularity.
A pencil that is fully extended cannot extend more. That is a singularity.
A fan that is at maximum speed cannot go faster. That is a singularity.
A borehole pump that is at maximum depth cannot go deeper. That is a singularity.
Singularity
Normal configuration:
Joint 1
/|
/ |
/ |
/ |
Joint 2
\ |
\ |
\ |
\|
End Effector
Singularity (fully stretched):
Joint 1
/|
/ |
/ |
/ |
Joint 2
|
|
|
End Effector
Cannot stretch further.
A singularity is a configuration where the robot loses some ability to move. Singularities cause problems in control.
Dynamics is the study of how forces affect motion. It looks at mass, inertia, gravity, and friction.
Dynamics tells us how much force is needed to move a robot. It tells us how heavy objects affect the robot. Without dynamics, the robot might use too much force or too little force.
Imagine pushing a car. If the car is heavy, you need more force. If the car is light, you need less force. Dynamics is about how much force is needed.
A car needs more force to go uphill than downhill. Dynamics explains why.
In physics class, you learn about Newton's laws. Force equals mass times acceleration. That is dynamics.
When you push a door, you use force. Dynamics describes how much force is needed.
A trader pushes a wheelbarrow full of goods. Dynamics describes how much force is needed.
Dynamics Force --> Mass --> Acceleration F = m * a Force = Mass times Acceleration
Dynamics is the study of how forces affect motion. It looks at mass, inertia, gravity, and friction.
Lagrangian dynamics is a way to calculate the forces needed to move a robot. It uses energy instead of forces directly.
Lagrangian dynamics is powerful. It can handle complex robots with many joints. It is used in advanced robot control.
Imagine a ball rolling down a hill. It has potential energy at the top. It has kinetic energy at the bottom. Lagrangian dynamics uses these energies to calculate motion.
A pendulum swings back and forth. Lagrangian dynamics describes its motion.
A roller coaster uses potential and kinetic energy. Lagrangian dynamics describes the motion.
A swing uses potential and kinetic energy. Lagrangian dynamics describes the motion.
A water wheel uses potential and kinetic energy. Lagrangian dynamics describes the motion.
Lagrangian Dynamics Potential Energy (PE) + Kinetic Energy (KE) L = KE - PE Calculate motion from L.
Lagrangian dynamics uses energy to calculate the forces needed to move a robot.
Newton-Euler dynamics is another way to calculate forces. It uses Newton's laws directly. It looks at each link of the robot one by one.
Newton-Euler dynamics is efficient for computers. It is used in real-time robot control.
Imagine a chain. Each link pulls the next link. Newton-Euler dynamics looks at each link and calculates the forces.
A train has many cars. Newton-Euler dynamics describes how each car pulls the next.
A tug-of-war rope has many people pulling. Newton-Euler dynamics describes the forces.
A chain on a bicycle has many links. Newton-Euler dynamics describes the forces.
A chain in a grinding machine has many links. Newton-Euler dynamics describes the forces.
Newton-Euler Dynamics Link 1 --> Link 2 --> Link 3 --> End Effector Calculate forces on each link.
Newton-Euler dynamics uses Newton's laws directly. It looks at each link of the robot one by one.
Wheeled mobile robot kinematics is the study of how wheeled robots move. It describes how wheel speeds affect the robot's motion.
Wheeled robots are common. They are used in factories, warehouses, and homes. Understanding their kinematics helps us control them.
Imagine a car. If both wheels turn at the same speed, the car goes straight. If one wheel turns faster, the car turns. That is wheeled mobile robot kinematics.
| Type | How It Moves | Example |
|---|---|---|
| Differential Drive | Two wheels, each with its own motor | Robot vacuum |
| Car-Like | Two wheels, steering wheel | Car |
| Tank-Like | Two tracks, each with its own motor | Tank |
| Omni-Directional | Special wheels that move in any direction | Special robots |
A car uses Car-Like kinematics. A robot vacuum uses Differential Drive.
A school robot might use Differential Drive.
A toy car might use Car-Like kinematics.
A keke napep uses Car-Like kinematics.
Differential Drive
Left Wheel Right Wheel
| |
V V
+-------+ +-------+
| Motor | | Motor |
+-------+ +-------+
| |
V V
+-------+ +-------+
| Wheel | | Wheel |
+-------+ +-------+
Both forward: go straight
Left faster: turn right
Right faster: turn left
Wheeled mobile robot kinematics describes how wheeled robots move. Different wheel types give different abilities.
Kinematic constraints are rules that limit how a robot can move. They describe what the robot cannot do.
Constraints affect robot control. A car cannot move sideways. A robot arm cannot bend backwards. Understanding constraints helps us design better controllers.
Imagine you are in a train. You can move forward and backward. You cannot move sideways. That is a kinematic constraint.
| Type | Meaning | Example |
|---|---|---|
| Holonomic | Can move in any direction | Omni-directional robot |
| Non-Holonomic | Cannot move sideways | Car |
A car cannot move sideways. That is a non-holonomic constraint.
A train can only move on tracks. That is a constraint.
A door can only swing. That is a constraint.
A danfo bus can only move on roads. That is a constraint.
Kinematic Constraints Holonomic: Can move anywhere +-------+ | Robot | +-------+ β β β β Non-Holonomic: Cannot move sideways +-------+ | Robot | +-------+ β β (no β β)
Kinematic constraints are rules that limit robot movement. They affect how we control robots.
Real robots use kinematics and dynamics to move and work.
| Robot | Kinematics | Purpose |
|---|---|---|
| Factory Robot Arm | Forward and inverse kinematics | Pick and place |
| Self-Driving Car | Car-Like kinematics | Drive on roads |
| Robot Vacuum | Differential drive | Clean floors |
| Surgical Robot | Precise kinematics | Perform surgery |
| Mars Rover | Rocky-Bogie kinematics | Explore Mars |
A surgical robot uses precise kinematics to move tiny instruments inside the body.
A school robot arm uses forward and inverse kinematics to pick up objects.
A robot vacuum uses differential drive kinematics to move around.
A robot in a Lagos factory uses kinematics to assemble products.
Real Robot: Surgical Robot
+-------------------+
| Robot Arm |
| +-----------+ |
| | Joint 1 | |
| +-----------+ |
| +-----------+ |
| | Joint 2 | |
| +-----------+ |
| +-----------+ |
| | Joint 3 | |
| +-----------+ |
+-------------------+
|
V
Precise movement for surgery
Real robots use kinematics and dynamics to move and work. Each robot's kinematics is suited for its job.
Debugging kinematic problems means finding and fixing issues in robot motion.
Kinematic problems can cause the robot to move incorrectly. It might miss its target. It might hit itself. It might get stuck. Debugging helps you fix these problems.
| Problem | Cause | Solution |
|---|---|---|
| Robot misses target | Wrong joint angles | Recalculate inverse kinematics |
| Robot hits itself | No collision checking | Add collision detection |
| Robot moves slowly | Joint velocity limits | Adjust velocity limits |
| Robot at singularity | Jacobian singular | Avoid singular configurations |
| Robot vibrates | Control loop instability | Tune controller gains |
If a robot arm misses its target, the inverse kinematics calculation may be wrong.
If a robot arm hits itself, add collision checking.
If a robot vacuum gets stuck, check its kinematics.
If a factory robot in Lagos misses its target, check the kinematic model.
Debugging Kinematic Problems
[ Robot misses target ]
|
V
[ Check joint angles ]
|
V
[ Check inverse kinematics ]
|
V
[ Check for singularities ]
|
V
[ Fix problem ]
|
V
[ Test again ]
Debugging kinematic problems means finding and fixing issues in robot motion. Common problems include missing targets, hitting itself, and singularities.
| Word | Simple Definition |
|---|---|
| Kinematics | The study of motion without forces. |
| Dynamics | The study of motion with forces. |
| Configuration Space | All possible positions of a robot. |
| Degrees of Freedom | The number of independent ways a robot can move. |
| Forward Kinematics | Finding hand position from joint angles. |
| Inverse Kinematics | Finding joint angles from hand position. |
| Jacobian Matrix | Relates joint velocities to hand velocities. |
| Singularity | A configuration where the robot loses some ability to move. |
| Lagrangian Dynamics | Using energy to calculate forces. |
| Newton-Euler Dynamics | Using Newton's laws to calculate forces. |
| Differential Drive | Two wheels, each with its own motor. |
| Kinematic Constraints | Rules that limit robot movement. |
| Holonomic | Can move in any direction. |
| Non-Holonomic | Cannot move sideways. |
| Concept | Real-Life Example |
|---|---|
| Kinematics | Throwing a ball describes its path. |
| Dynamics | A car needs more force uphill. |
| Configuration Space | A door can be at many angles. |
| Degrees of Freedom | Your shoulder has 3 DOF. |
| Forward Kinematics | A crane operator knows where the hook is. |
| Inverse Kinematics | Your brain calculates arm angles to touch a spot. |
| Jacobian Matrix | A bicycle's pedaling speed affects wheel speed. |
| Singularity | A fully stretched arm cannot stretch more. |
| Lagrangian Dynamics | A pendulum swings using energy. |
| Newton-Euler Dynamics | A train has many cars pulling each other. |
| Wheeled Kinematics | A car turns by steering. |
| Kinematic Constraints | A car cannot move sideways. |
| Concept | Nigerian Example |
|---|---|
| Kinematics | A danfo bus travels from Yaba to Ikeja. |
| Dynamics | A trader pushes a wheelbarrow full of goods. |
| Configuration Space | A traffic warden can point in many directions. |
| Degrees of Freedom | A keke napep has 2 DOF. |
| Forward Kinematics | A borehole pipe's angle determines where water comes out. |
| Inverse Kinematics | A farmer calculates arm angles to reach fruit. |
| Jacobian Matrix | A mortar and pestle relates movement to grinding. |
| Singularity | A borehole pump at maximum depth cannot go deeper. |
| Lagrangian Dynamics | A water wheel uses potential and kinetic energy. |
| Newton-Euler Dynamics | A grinding machine chain pulls link by link. |
| Wheeled Kinematics | A keke napep uses Car-Like kinematics. |
| Kinematic Constraints | A danfo bus can only move on roads. |
| Concept | Everyday Example |
|---|---|
| Kinematics | Walking from one room to another. |
| Dynamics | Pushing a shopping cart. |
| Configuration Space | A fan rotating to different angles. |
| Degrees of Freedom | A pair of scissors opens and closes. |
| Forward Kinematics | Knowing where the tip of a pencil is. |
| Inverse Kinematics | Reaching for a cup. |
| Jacobian Matrix | A can opener handle and blade. |
| Singularity | A door fully open. |
| Lagrangian Dynamics | A swing moving. |
| Newton-Euler Dynamics | A chain on a bicycle. |
| Wheeled Kinematics | A car turning. |
| Kinematic Constraints | A door only swings. |
| Mistake | Why It Is Wrong | How to Fix It |
|---|---|---|
| Confusing kinematics and dynamics | They are different concepts. | Remember: kinematics = motion, dynamics = forces. |
| Ignoring singularities | Robot can become unstable. | Check for singularities in path planning. |
| Wrong inverse kinematics | Robot misses target. | Check calculations and multiple solutions. |
| Forgetting constraints | Robot cannot move as planned. | Consider kinematic constraints. |
| Not checking collisions | Robot hits itself or obstacles. | Add collision detection. |
| Ignoring dynamics | Robot uses wrong force. | Calculate dynamic forces. |
Forward Kinematics:
Joint angles --> Hand position
Inverse Kinematics:
Hand position --> Joint angles
+-------------+ +-------------+
| Joint angles| --> | Hand position|
+-------------+ +-------------+
^ |
| |
+-------------------+
Inverse Kinematics
( Start )
|
V
+----------------+
| Desired |
| position |
+----------------+
|
V
+----------------+
| Set up |
| equations |
+----------------+
|
V
+----------------+
| Solve for |
| joint angles |
+----------------+
|
V
+----------------+
| Multiple |
| solutions? |
+----------------+
/ \
YES NO
/ \
V V
+---------+ +-----------+
| Choose | | Use |
| best | | solution |
+---------+ +-----------+
\ /
\ /
V V
+----------------+
| Check for |
| singularities |
+----------------+
|
V
+----------------+
| Move robot |
+----------------+
| Aspect | Kinematics | Dynamics |
|---|---|---|
| Studies | Motion | Forces |
| Looks at | Position, velocity, acceleration | Force, mass, inertia |
| Example | Where is the hand? | How much force to lift? |
| Tools | Jacobian, forward/inverse kinematics | Lagrangian, Newton-Euler |
Step 1: Define target position
|
V
Step 2: Solve inverse kinematics
|
V
Step 3: Check for singularities
|
V
Step 4: Calculate dynamics
|
V
Step 5: Plan trajectory
|
V
Step 6: Execute motion
|
V
Step 7: Verify
Kinematics is the study of motion. It looks at position, velocity, and acceleration.
Dynamics is the study of motion with forces. It looks at how forces affect movement.
Configuration space describes all possible positions of a robot.
Degrees of freedom is the number of independent ways a robot can move.
Forward kinematics finds the position of the robot's hand from joint angles.
Inverse kinematics finds joint angles from the position of the robot's hand.
The Jacobian matrix relates joint velocities to end-effector velocities.
A singularity is a configuration where the robot loses some ability to move.
Dynamics is the study of how forces affect motion.
Lagrangian dynamics uses energy to calculate the forces needed to move a robot.
Newton-Euler dynamics uses Newton's laws directly, looking at each link.
Wheeled mobile robot kinematics describes how wheeled robots move.
Kinematic constraints are rules that limit how a robot can move.
Real robots use kinematics and dynamics to move and work.
Debugging kinematic problems means finding and fixing issues in robot motion.
In this module, you learned about advanced kinematics and dynamics. You learned that kinematics is the study of motion without forces. You learned that dynamics is the study of motion with forces.
You learned about configuration space, degrees of freedom, forward kinematics, and inverse kinematics. You learned about the Jacobian matrix and singularities. You learned about Lagrangian dynamics and Newton-Euler dynamics. You learned about wheeled mobile robot kinematics and kinematic constraints.
You learned how to solve forward and inverse kinematics problems. You learned how to debug kinematic problems. You learned about real robots and their kinematics.
Most importantly, you learned that kinematics and dynamics are the foundation of advanced robotics. They are used in factory robots, surgical robots, self-driving cars, and space robots.
In the next module, you will learn about Probabilistic Robotics. You will learn how robots deal with uncertainty. You will learn about Bayesian filtering, Kalman filters, particle filters, and SLAM.
But for now, take a moment to celebrate what you have learned. You have taken another big step in your journey to becoming a robotics expert. Well done!
Match the term on the left with its definition on the right.
| Term | Definition |
|---|---|
| 1. Kinematics | A. The study of motion with forces |
| 2. Dynamics | B. Finding hand position from joint angles |
| 3. Forward Kinematics | C. The study of motion without forces |
| 4. Inverse Kinematics | D. A configuration where the robot loses ability to move |
| 5. Jacobian Matrix | E. Finding joint angles from hand position |
| 6. Singularity | F. Relates joint velocities to hand velocities |
| 7. Lagrangian Dynamics | G. Uses Newton's laws directly |
| 8. Newton-Euler Dynamics | H. Uses energy to calculate forces |
Answers: 1-C, 2-A, 3-B, 4-E, 5-F, 6-D, 7-H, 8-G
Title: Solve a Forward Kinematics Problem
Instructions:
Example:
Link 1: 10 cm Link 2: 8 cm Joint 1 angle: 30 degrees Joint 2 angle: 45 degrees Calculate end effector position.
Title: Kinematics Scavenger Hunt
Instructions:
Example:
| Object | Degrees of Freedom | Movement |
|---|---|---|
| Door | 1 | Swings open/closed |
| Scissors | 1 | Opens/closes |
| Fan | 1 | Rotates |
| Car | 2 | Forward/backward, turn |
| Robot arm | 3 | Reach in 3D space |
Title: Build a Simple Robot Arm
Goal: Create a 2-joint robot arm and calculate its kinematics.
Steps:
Deliverables:
Title: Implement Inverse Kinematics in Simulation
Instructions:
Grading Criteria:
| Criteria | Points |
|---|---|
| Robot reaches target | 30 |
| Inverse kinematics is correct | 25 |
| Multiple targets tested | 15 |
| Report is clear | 15 |
| Singularities avoided | 15 |
| Total | 100 |
In Module Two, you will learn about Probabilistic Robotics. You will learn how robots deal with uncertainty.
You will learn about:
To prepare for Module Two:
Get ready for an exciting journey into the world of probabilistic robotics!
Congratulations! You have completed Module One of Fundamentals of Robotics Level Three. You have learned about advanced kinematics and dynamics.
You learned that kinematics is the study of motion without forces. You learned that dynamics is the study of motion with forces. You learned about configuration space and degrees of freedom. You learned about forward kinematics and inverse kinematics.
You learned about the Jacobian matrix and singularities. You learned about Lagrangian dynamics and Newton-Euler dynamics. You learned about wheeled mobile robot kinematics and kinematic constraints. You learned how to solve forward and inverse kinematics problems. You learned how to debug kinematic problems. You learned about real robots and their kinematics.
You also learned many examples from Nigeria, from your home, from school, and from everyday life. You learned through stories, illustrations, and activities.
Now you are ready for Module Two: Probabilistic Robotics. In Module Two, you will learn how robots deal with uncertainty. You will learn about Bayesian filtering, Kalman filters, particle filters, and SLAM.
But before you move on, take a moment to review this module. Make sure you understand the key ideas. Practise solving kinematics problems. Draw diagrams. Test your calculations. The more you practise, the better you will become.
You are doing great. Keep learning. Keep exploring. Keep building. The world of robotics is waiting for you!
End of Module One
Next: Module Two β Probabilistic Robotics
Welcome to Module Two of Level Three! In Module One, you learned about advanced kinematics and dynamics. You learned how robots move and how forces affect their motion. Now you will learn about something even more important: how robots deal with uncertainty.
In the real world, nothing is perfect. Sensors give noisy readings. Motors slip. The ground is uneven. The robot never knows exactly where it is. It only has a guess β a belief β about its position.
Probabilistic robotics is the study of how robots make decisions when they are not sure. It uses probability β the mathematics of chance β to represent what the robot knows.
Think about it this way. When you walk into a dark room, you are not sure where the furniture is. But you have a rough idea. You use your hands to feel around. You update your belief with every step. Robots do the same thing with sensors and mathematics.
In this module, you will learn about Bayesian filtering, Kalman filters, particle filters, and SLAM. These are the tools that make self-driving cars, robot vacuums, and Mars rovers possible.
Do not worry if these words sound hard. We will explain everything step by step, using simple examples and stories.
Let us begin!
By the end of this module, you will be able to:
Once upon a time, in the city of Ibadan, Nigeria, there lived a boy named Chidi. Chidi had built a small robot named Bola. Bola had wheels, motors, sensors, and a small computer brain. Chidi had programmed Bola to move around the house.
One day, Chidi's mother asked him to bring her phone from the bedroom. Chidi decided to send Bola to get it. He programmed Bola to go to the bedroom, pick up the phone, and come back.
But there was a problem. Bola did not know exactly where she was. Her wheels slipped a little when she moved. Her sensors gave readings that were not always accurate. The floor was slightly uneven. Bola was not sure of her position.
Bola started moving. She went forward. She turned left. She went forward again. But because of the wheel slip, she was not where she thought she was. She bumped into a chair.
Chidi watched in frustration. "Bola, you are in the wrong place!" he said.
Bola did not answer. She just kept moving.
Chidi went to his father, who was an engineer. "Papa, Bola does not know where she is. How can I help her?"
His father smiled. "Chidi, this is a problem of uncertainty. Bola never knows exactly where she is. She only has a guess. You need to teach her to use probability."
"What is probability?" Chidi asked.
"Probability is the mathematics of chance," his father explained. "It tells you how likely something is. Bola can use probability to keep track of where she might be."
Chidi was fascinated. He learned about beliefs β the robot's guess about its position. He learned about Bayesian filtering β a way to update beliefs when new information comes in. He learned about particle filters β a way to represent many possible positions.
He wrote a new program for Bola. The program kept track of many possible positions. Every time Bola moved, the program updated the positions. Every time Bola's sensors detected something, the program updated again.
He tested the new program. Bola moved forward. She turned left. She went forward again. This time, she knew roughly where she was. She avoided the chair. She reached the bedroom. She picked up the phone. She came back.
Chidi's mother was amazed. "Bola is so smart!" she said.
Chidi smiled. He had taught Bola to deal with uncertainty. He had taught her probabilistic robotics.
That is what you will learn in this module. You will learn how robots deal with uncertainty. You will learn about probability, beliefs, Bayesian filtering, Kalman filters, particle filters, and SLAM.
Let us begin!
Uncertainty means not being sure. In robotics, uncertainty means the robot does not know exactly what is happening.
Robots never have perfect information. Sensors have noise. Motors slip. The environment changes. If a robot ignores uncertainty, it will make mistakes. If it understands uncertainty, it can make better decisions.
Imagine you are in a dark room. You are not sure where the furniture is. You have some idea, but you are not certain. That is uncertainty.
When you look at a clock from far away, you might not be sure of the exact time. That is uncertainty.
When you guess the answer to a question you are not sure about, that is uncertainty.
When you are not sure if the food is salty enough, that is uncertainty.
When a driver is not sure if the road ahead is clear, that is uncertainty.
Uncertainty Robot's belief: "I am somewhere here" +-------------------+ | ? ? ? ? ? ? | | ? ? ? ? ? ? | | ? ? ? ? ? ? | +-------------------+ The robot is not sure exactly where it is.
Uncertainty means not being sure. Robots always face uncertainty because sensors and motors are not perfect.
Probability is the mathematics of chance. It tells you how likely something is. It is measured between 0 and 1.
Probability helps robots represent uncertainty. Instead of saying "I am here," the robot says "I am 80% sure I am here." That is more useful.
Imagine flipping a coin. There are two possible outcomes: heads or tails. Each has a probability of 0.5, or 50%. That means there is an equal chance of either outcome.
The weather forecast says "70% chance of rain." That is probability.
If you guess the answer to a multiple-choice question with four options, you have a 25% chance of being right.
If you are not sure whether the food is ready, you might say "I am 60% sure." That is probability.
A trader might say "There is a 90% chance I will sell all my goods today." That is probability.
Probability Scale 0% 25% 50% 75% 100% |---------|---------|---------|---------| Impossible Unlikely Maybe Likely Certain
Probability is the mathematics of chance. It tells you how likely something is. It is measured between 0 and 1.
A belief is what a robot thinks is true. It is the robot's guess about the world. It is based on probability.
A robot cannot know everything for certain. It only has beliefs. Beliefs help the robot make decisions. They are updated as new information comes in.
Imagine you are looking for your shoes. You think they are in your room, but you are not sure. That is your belief. When you find them, your belief changes.
You believe your keys are in your bag. When you check and find them, your belief is confirmed.
You believe the test is on Friday. When the teacher confirms it, your belief is confirmed.
You believe there is milk in the fridge. When you check and find none, your belief changes.
A trader believes there will be many customers at the market. When she arrives and sees few people, her belief changes.
Belief Before: +-------------+ | I think I | | am here: | | (x=5, y=3) | +-------------+ After moving: +-------------+ | I think I | | am here: | | (x=7, y=3) | +-------------+ Beliefs change as the robot moves.
A belief is what a robot thinks is true. It is the robot's guess about the world. Beliefs are updated as new information comes in.
Bayesian filtering is a way to update beliefs when new information comes in. It uses Bayes' theorem to combine what the robot already knows with new sensor readings.
Robots always get new information. Sensors give readings. Motors move the robot. Bayesian filtering helps the robot update its belief with each new piece of information.
Imagine you are looking for your friend in a crowd. You think he is on the left side. Then you see a red shirt on the right side. You update your belief. Maybe he moved. That is Bayesian filtering.
Weather forecasters update their predictions as new data comes in. That is Bayesian filtering.
You guess the answer to a question. Then you remember something from class. You update your answer. That is Bayesian filtering.
You think the food needs more salt. You taste it again. You update your belief. That is Bayesian filtering.
A trader thinks a customer will buy. The customer smiles. The trader updates her belief. That is Bayesian filtering.
Bayesian Filtering
Prior belief: "I am here"
|
V
New sensor reading: "I see a wall"
|
V
Updated belief: "I am here, near a wall"
Belief + New information = Updated belief
Bayesian filtering is a way to update beliefs when new information comes in. It uses Bayes' theorem to combine what the robot already knows with new sensor readings.
A Kalman filter is a special type of Bayesian filter. It is used when the robot's belief can be represented by a single guess with some uncertainty.
Kalman filters are fast and accurate. They are used in many real robots, including self-driving cars and drones.
Imagine you are driving a car. You think you are going 60 km/h. The speedometer says 62 km/h. You combine both to get a better estimate. That is a Kalman filter.
Your phone's GPS uses a Kalman filter to estimate your position.
A science experiment might use a Kalman filter to estimate the temperature.
A thermostat uses a Kalman filter to estimate the room temperature.
A generator's control system uses a Kalman filter to estimate the engine speed.
Kalman Filter
Predicted position: (10, 5) Β± 2 cm
Sensor reading: (11, 5) Β± 1 cm
|
V
Updated position: (10.7, 5) Β± 0.8 cm
Combines prediction and sensor.
A Kalman filter is a special type of Bayesian filter. It is used when the robot's belief can be represented by a single guess with some uncertainty.
A particle filter is a type of Bayesian filter that uses many "particles" to represent the robot's belief. Each particle is a guess about the robot's position.
Particle filters can handle complex beliefs. They can represent many possible positions at once. They are used in robot localization and SLAM.
Imagine you are lost in a city. You do not know where you are. You make many guesses: "Maybe I am here. Maybe I am there. Maybe I am over there." Each guess is a particle. As you get new information, you discard bad guesses and keep good ones.
Search and rescue teams use particle filters to track missing people.
A school project might use a particle filter to track a robot in a maze.
A robot vacuum uses a particle filter to track its position in your house.
A delivery robot in Lagos uses a particle filter to navigate crowded streets.
Particle Filter Initial belief: many particles +-------------------+ | ? ? ? ? ? ? ? ? | | ? ? ? ? ? ? ? ? | | ? ? ? ? ? ? ? ? | +-------------------+ After sensor update: fewer particles +-------------------+ | ? ? ? | | ? ? ? | | ? ? ? | +-------------------+ After more updates: particles converge +-------------------+ | ? | | ? | | ? | +-------------------+
A particle filter is a type of Bayesian filter that uses many particles to represent the robot's belief. Each particle is a guess about the robot's position.
Localization is the process of figuring out where the robot is. It uses sensors, motion, and a map to estimate the robot's position.
Without localization, a robot cannot navigate. It cannot plan paths. It cannot reach its destination. Localization is the foundation of autonomous navigation.
Imagine you wake up in a new house. You look around. You see a kitchen. You see a bedroom. You figure out where you are. That is localization.
Your phone's GPS uses localization to show you where you are on the map.
A robot in a maze uses localization to figure out which part of the maze it is in.
A robot vacuum uses localization to know which room it is cleaning.
A danfo driver uses localization to know which bus stop he is at.
Localization
Map:
+-------------------+
| +---+ +---+ |
| | A | | B | |
| +---+ +---+ |
| |
| +---+ +---+ |
| | C | | D | |
| +---+ +---+ |
+-------------------+
Robot: "I see a red door and a window."
|
V
Robot: "I must be in room A."
Localization is the process of figuring out where the robot is. It uses sensors, motion, and a map to estimate the robot's position.
A map is a representation of the environment. It tells the robot where things are.
Without a map, the robot cannot plan paths. It cannot avoid obstacles. Maps help robots understand their environment.
| Type | Description | Example |
|---|---|---|
| Occupancy Grid | Divides space into cells, each marked free or occupied | Robot vacuum map |
| Feature Map | Stores landmarks like doors and walls | Self-driving car map |
| Topological Map | Stores connections between places | Subway map |
| Metric Map | Stores exact distances and angles | CAD map |
Google Maps is a map of the world.
A school map shows where classrooms are.
A floor plan shows where rooms are in a house.
A market map shows where different stalls are.
Occupancy Grid Map +---+---+---+---+ | 0 | 0 | 1 | 0 | +---+---+---+---+ | 0 | 1 | 1 | 0 | +---+---+---+---+ | 0 | 0 | 0 | 0 | +---+---+---+---+ 0 = free space 1 = occupied
A map is a representation of the environment. Different types of maps are used for different purposes.
SLAM stands for Simultaneous Localization and Mapping. It means building a map and finding your position at the same time.
Sometimes a robot does not have a map. It must build one as it moves. SLAM lets the robot explore unknown places and create a map.
Imagine exploring a new city. You do not have a map. As you walk, you draw a map in your mind. You also remember where you are on that map. That is SLAM.
A robot vacuum uses SLAM to map your house as it cleans.
A robot in a science fair might use SLAM to explore a maze.
A robot toy might use SLAM to explore your house.
A robot used in a new building in Abuja might use SLAM to map the building.
SLAM Process
[ Start with no map ]
|
V
[ Move and sense ]
|
V
[ Build map ]
|
V
[ Find position ]
|
V
[ Update map ]
|
V
[ Repeat ]
SLAM means building a map and finding your position at the same time. It lets robots explore unknown places.
A sensor model is a mathematical description of how a sensor behaves. It tells the robot how likely a sensor reading is, given the robot's position.
Sensors are not perfect. They have noise. They sometimes give wrong readings. A sensor model helps the robot understand how much to trust the sensor.
Imagine you are measuring a table with a ruler. The ruler might be slightly off. A sensor model tells you how much error to expect.
A thermometer might read 25Β°C when the actual temperature is 24.8Β°C. A sensor model describes this error.
A science experiment might use a sensor model to account for measurement error.
A bathroom scale might be slightly inaccurate. A sensor model describes the error.
A trader's weighing scale might be slightly off. A sensor model describes the error.
Sensor Model Actual distance: 10 cm Sensor reading: 10.2 cm (noise +0.2) Sensor model: P(reading | actual distance) Tells how likely a reading is.
A sensor model is a mathematical description of how a sensor behaves. It tells the robot how much to trust the sensor.
A motion model is a mathematical description of how the robot moves. It tells the robot how likely a movement is, given the robot's controls.
Robots do not move perfectly. Wheels slip. Motors have errors. A motion model helps the robot understand how much to trust its movement.
Imagine you tell a robot to move forward 10 cm. It might move 9.8 cm or 10.2 cm. A motion model describes this error.
A car's odometer might be slightly inaccurate. A motion model describes the error.
A robot in a maze might slip on a smooth floor. A motion model describes this.
A robot vacuum might slip on a rug. A motion model describes this.
A delivery robot might slip on a wet road. A motion model describes this.
Motion Model Command: move forward 10 cm Actual movement: 9.8 cm to 10.2 cm Motion model: P(new position | old position, command) Tells how likely a movement is.
A motion model is a mathematical description of how the robot moves. It tells the robot how much to trust its movement.
The Bayes filter algorithm is a step-by-step method for updating beliefs. It has two steps: prediction and update.
The Bayes filter is the foundation of probabilistic robotics. Kalman filters and particle filters are special cases of the Bayes filter.
Imagine you are tracking a friend in a crowd. Step 1: Predict where he will go. Step 2: Look around and update your prediction. That is the Bayes filter.
A weather forecaster predicts rain, then updates with new data.
You predict your exam score, then update after seeing the questions.
You predict the food is ready, then update after tasting it.
A trader predicts sales, then updates after seeing customers.
Bayes Filter Algorithm
Belief at time t-1
|
V
Prediction step (motion model)
|
V
Predicted belief
|
V
Update step (sensor model)
|
V
Updated belief at time t
|
V
Repeat
The Bayes filter algorithm is a step-by-step method for updating beliefs. It has two steps: prediction and update.
Localization with particle filters means using many particles to estimate the robot's position.
Particle filters can handle complex beliefs. They can represent many possible positions at once. They are widely used in robotics.
Imagine you are lost in a city. You make many guesses about where you are. Each guess is a particle. As you move and see things, you discard bad guesses and keep good ones.
A robot vacuum uses particle filter localization to track its position.
A robot in a maze uses particle filter localization to find its way.
A robot toy uses particle filter localization to follow you.
A delivery robot in Lagos uses particle filter localization to navigate.
Particle Filter Localization Step 1: Initialize +-------------------+ | ? ? ? ? ? ? ? ? | | ? ? ? ? ? ? ? ? | | ? ? ? ? ? ? ? ? | +-------------------+ Step 2: Predict +-------------------+ | ? ? ? ? ? ? ? | | ? ? ? ? ? ? | | ? ? ? ? ? | +-------------------+ Step 3: Update +-------------------+ | ? ? ? | | ? ? ? | | ? ? ? | +-------------------+ Step 4: Resample +-------------------+ | ? | | ? | | ? | +-------------------+
Localization with particle filters means using many particles to estimate the robot's position.
Real robots use probabilistic robotics to deal with uncertainty.
| Robot | Probabilistic Method | Purpose |
|---|---|---|
| Robot Vacuum | Particle filter SLAM | Map and clean floors |
| Self-Driving Car | Kalman filter, particle filter | Navigate roads |
| Mars Rover | Bayesian filtering | Explore Mars |
| Surgical Robot | Kalman filter | Perform surgery |
| Drone | Kalman filter | Fly stably |
A self-driving car uses Kalman filters to estimate its position and speed.
A school robot uses particle filters to navigate a maze.
A robot vacuum uses SLAM to map your house.
A robot in a Lagos warehouse uses probabilistic methods to move goods.
Real Robot: Self-Driving Car
+-------------------+
| Sensors |
| (GPS, camera, |
| radar) |
+-------------------+
|
V
+-------------------+
| Kalman Filter |
| (estimates |
| position) |
+-------------------+
|
V
+-------------------+
| Path Planning |
+-------------------+
|
V
+-------------------+
| Control |
+-------------------+
Real robots use probabilistic robotics to deal with uncertainty. Kalman filters, particle filters, and SLAM are common methods.
Debugging probabilistic systems means finding and fixing problems in systems that use probability.
Probabilistic systems can fail. The robot might get lost. The belief might become wrong. Debugging helps you find and fix these problems.
| Problem | Cause | Solution |
|---|---|---|
| Robot gets lost | Particles all in wrong place | Increase particles or improve sensor model |
| Belief does not converge | Sensor model is wrong | Recalibrate sensor model |
| Robot is overconfident | Noise not modelled | Add noise to motion and sensor models |
| Robot is underconfident | Too much noise | Reduce noise in models |
| SLAM map is wrong | Loop closure error | Add loop closure detection |
If your phone's GPS is inaccurate, the sensor model might be wrong.
If your robot gets lost in a maze, check the particle filter.
If your robot vacuum gets stuck, check the SLAM map.
If a delivery robot in Lagos gets lost, check the localization system.
Debugging Probabilistic Systems
[ Robot gets lost ]
|
V
[ Check particles ]
|
V
[ Check sensor model ]
|
V
[ Check motion model ]
|
V
[ Fix problem ]
|
V
[ Test again ]
Debugging probabilistic systems means finding and fixing problems in systems that use probability. Common problems include getting lost, not converging, and wrong maps.
| Word | Simple Definition |
|---|---|
| Uncertainty | Not being sure. |
| Probability | The mathematics of chance. |
| Belief | What a robot thinks is true. |
| Bayesian Filtering | A way to update beliefs when new information comes in. |
| Kalman Filter | A type of Bayesian filter for single guesses. |
| Particle Filter | A type of Bayesian filter using many particles. |
| Localization | Figuring out where the robot is. |
| Map | A representation of the environment. |
| SLAM | Simultaneous Localization and Mapping. |
| Sensor Model | Describes how a sensor behaves. |
| Motion Model | Describes how the robot moves. |
| Occupancy Grid | A map that divides space into cells. |
| Concept | Real-Life Example |
|---|---|
| Uncertainty | Looking at a clock from far away. |
| Probability | Weather forecast says 70% chance of rain. |
| Belief | Thinking your keys are in your bag. |
| Bayesian Filtering | Weather forecasters update predictions. |
| Kalman Filter | Phone's GPS estimates position. |
| Particle Filter | Search and rescue teams track missing people. |
| Localization | Phone's GPS shows you where you are. |
| Map | Google Maps is a map of the world. |
| SLAM | Robot vacuum maps your house. |
| Sensor Model | Thermometer reads 25Β°C when actual is 24.8Β°C. |
| Motion Model | Car odometer slightly inaccurate. |
| Bayes Filter | Weather forecaster predicts and updates. |
| Concept | Nigerian Example |
|---|---|
| Uncertainty | A driver not sure if the road ahead is clear. |
| Probability | A trader says 90% chance of selling all goods. |
| Belief | A trader believes there will be many customers. |
| Bayesian Filtering | A trader updates belief when a customer smiles. |
| Kalman Filter | A generator's control system estimates speed. |
| Particle Filter | A delivery robot in Lagos navigates streets. |
| Localization | A danfo driver knows which bus stop he is at. |
| Map | A market map shows where stalls are. |
| SLAM | A robot maps a new building in Abuja. |
| Sensor Model | A trader's scale is slightly off. |
| Motion Model | A delivery robot slips on a wet road. |
| Bayes Filter | A trader predicts sales, updates with customers. |
| Concept | Everyday Example |
|---|---|
| Uncertainty | Not sure if the food is salty enough. |
| Probability | Weather forecast says 70% chance of rain. |
| Belief | Thinking there is milk in the fridge. |
| Bayesian Filtering | Tasting food and updating your belief. |
| Kalman Filter | Thermostat estimates room temperature. |
| Particle Filter | Robot vacuum tracks position. |
| Localization | Robot vacuum knows which room it is in. |
| Map | A floor plan of your house. |
| SLAM | Robot toy explores your house. |
| Sensor Model | Bathroom scale is slightly inaccurate. |
| Motion Model | Robot vacuum slips on a rug. |
| Bayes Filter | Predicting food is ready, then tasting it. |
| Mistake | Why It Is Wrong | How to Fix It |
|---|---|---|
| Ignoring uncertainty | Robot makes wrong decisions. | Use probability to represent uncertainty. |
| Using too few particles | Belief is inaccurate. | Increase the number of particles. |
| Wrong sensor model | Belief does not converge. | Recalibrate sensor model. |
| Wrong motion model | Belief drifts away. | Recalibrate motion model. |
| Ignoring loop closure | SLAM map is wrong. | Add loop closure detection. |
| Not debugging | Problems continue. | Find and fix problems. |
+-------------------+
| Sensors |
| (noisy data) |
+-------------------+
|
V
+-------------------+
| Bayesian Filter |
| (updates belief)|
+-------------------+
|
V
+-------------------+
| Belief |
| (robot's guess) |
+-------------------+
|
V
+-------------------+
| Decision |
+-------------------+
|
V
+-------------------+
| Action |
+-------------------+
|
| (feedback)
V
+-------------------+
| Motion Model |
+-------------------+
( Start )
|
V
+----------------+
| Initialize |
| particles |
+----------------+
|
V
+----------------+
| Predict |
| (move) |
+----------------+
|
V
+----------------+
| Update |
| (weight) |
+----------------+
|
V
+----------------+
| Resample |
+----------------+
|
V
+----------------+
| Estimate |
| position |
+----------------+
|
V
( Repeat )
| Filter | Belief Representation | Speed | Flexibility |
|---|---|---|---|
| Kalman Filter | Single Gaussian | Fast | Low |
| Extended Kalman Filter | Single Gaussian (linearized) | Fast | Medium |
| Particle Filter | Many particles | Slower | High |
| Histogram Filter | Grid of probabilities | Medium | Medium |
Step 1: Initialize map
|
V
Step 2: Move robot
|
V
Step 3: Sense environment
|
V
Step 4: Update position
|
V
Step 5: Update map
|
V
Step 6: Repeat
|
V
Step 7: Loop closure
Uncertainty means not being sure. Robots always face uncertainty because sensors and motors are not perfect.
Probability is the mathematics of chance. It tells you how likely something is.
A belief is what a robot thinks is true. It is the robot's guess about the world.
Bayesian filtering is a way to update beliefs when new information comes in.
A Kalman filter is a special type of Bayesian filter used when the belief can be represented by a single guess.
A particle filter is a type of Bayesian filter that uses many particles to represent the robot's belief.
Localization is the process of figuring out where the robot is.
A map is a representation of the environment. Different types of maps are used for different purposes.
SLAM means building a map and finding your position at the same time.
A sensor model describes how a sensor behaves and how much to trust it.
A motion model describes how the robot moves and how much to trust its movement.
The Bayes filter algorithm is a step-by-step method for updating beliefs with prediction and update steps.
Localization with particle filters uses many particles to estimate the robot's position.
Real robots use probabilistic robotics to deal with uncertainty.
Debugging probabilistic systems means finding and fixing problems in systems that use probability.
In this module, you learned about probabilistic robotics. You learned that uncertainty is a fundamental part of robotics because sensors and motors are not perfect.
You learned about probability β the mathematics of chance. You learned about beliefs β the robot's guesses about the world. You learned about Bayesian filtering β a way to update beliefs when new information comes in.
You learned about Kalman filters and particle filters. You learned about localization β figuring out where the robot is. You learned about maps and map representation. You learned about SLAM β building a map and finding your position at the same time.
You learned about sensor models and motion models. You learned about the Bayes filter algorithm. You learned about localization with particle filters. You learned about real robots and their probabilistic methods. You learned about debugging probabilistic systems.
Most importantly, you learned that probabilistic robotics is what allows robots to work in the real world, where nothing is perfect.
In the next module, you will learn about Motion Planning. You will learn how robots plan paths through complex environments. You will learn about RRT, RRT*, trajectory optimization, and more.
But for now, take a moment to celebrate what you have learned. You have taken another big step in your journey to becoming a robotics expert. Well done!
Match the term on the left with its definition on the right.
| Term | Definition |
|---|---|
| 1. Uncertainty | A. The mathematics of chance |
| 2. Probability | B. What a robot thinks is true |
| 3. Belief | C. Not being sure |
| 4. Bayesian Filtering | D. Figuring out where the robot is |
| 5. Kalman Filter | E. A type of Bayesian filter for single guesses |
| 6. Particle Filter | F. Updating beliefs with new information |
| 7. Localization | G. A type of Bayesian filter using many particles |
| 8. SLAM | H. Building a map and finding position together |
Answers: 1-C, 2-A, 3-B, 4-F, 5-E, 6-G, 7-D, 8-H
Title: Particle Filter Localization Game
Instructions:
Example:
Robot: "I see a door on my left." Particles: Move and update based on this information. Robot: "I hear a clock ticking." Particles: Update again. Repeat until particles converge.
Title: Probability Scavenger Hunt
Instructions:
Example:
| Thing | What Is Uncertain | Probability |
|---|---|---|
| Weather | Will it rain? | 70% chance |
| Traffic | Will the road be clear? | 50% chance |
| Exam | Will I pass? | 80% chance |
| Food | Is it ready? | 60% chance |
| Bus | Will it arrive on time? | 40% chance |
Title: Build a Simple Localization System
Goal: Create a program that estimates a robot's position using probability.
Steps:
Deliverables:
Title: Implement a Particle Filter in Simulation
Instructions:
Grading Criteria:
| Criteria | Points |
|---|---|
| Particle filter works correctly | 30 |
| Particles converge to correct position | 25 |
| Multiple particle counts tested | 15 |
| Report is clear | 15 |
| Visualization is clear | 15 |
| Total | 100 |
In Module Three, you will learn about Motion Planning. You will learn how robots plan paths through complex environments.
You will learn about:
To prepare for Module Three:
Get ready for an exciting journey into the world of motion planning!
Congratulations! You have completed Module Two of Fundamentals of Robotics Level Three. You have learned about probabilistic robotics.
You learned that uncertainty is a fundamental part of robotics. You learned about probability and beliefs. You learned about Bayesian filtering, Kalman filters, and particle filters. You learned about localization, maps, and SLAM. You learned about sensor models and motion models. You learned about the Bayes filter algorithm. You learned about localization with particle filters. You learned about real robots and their probabilistic methods. You learned about debugging probabilistic systems.
You also learned many examples from Nigeria, from your home, from school, and from everyday life. You learned through stories, illustrations, and activities.
Now you are ready for Module Three: Motion Planning. In Module Three, you will learn how robots plan paths through complex environments. You will learn about configuration space, cell decomposition, roadmap methods, RRT, RRT*, and trajectory optimization.
But before you move on, take a moment to review this module. Make sure you understand the key ideas. Practise probability problems. Draw diagrams. Test your understanding. The more you practise, the better you will become.
You are doing great. Keep learning. Keep exploring. Keep building. The world of robotics is waiting for you!
End of Module Two
Next: Module Three β Motion Planning
Welcome to Module Three of Level Three! In Module One, you learned about advanced kinematics and dynamics. In Module Two, you learned about probabilistic robotics β how robots deal with uncertainty. Now you will learn about motion planning.
Motion planning is the art of figuring out how to move a robot from one place to another without hitting anything. It is like solving a puzzle. You have a start point. You have a goal. You have obstacles in between. You need to find a path that works.
Think about walking through a crowded market. You want to get from one end to the other. There are people, stalls, and potholes. You look ahead. You plan your steps. You avoid obstacles. That is motion planning.
Robots need motion planning too. A robot arm needs to move from one position to another without hitting itself. A self-driving car needs to find a route through traffic. A drone needs to fly around buildings. A robot vacuum needs to clean a room without getting stuck.
In this module, you will learn about configuration space, cell decomposition, roadmap methods, RRT, RRT*, and trajectory optimization. These are the tools that make robot motion possible.
Do not worry if these words sound hard. We will explain everything step by step, using simple examples and stories.
Let us begin!
By the end of this module, you will be able to:
Once upon a time, in the city of Enugu, Nigeria, there lived a girl named Ngozi. Ngozi had built a small robot named Musa. Musa had wheels, motors, sensors, and a small computer brain. Ngozi had programmed Musa to navigate around obstacles.
One day, Ngozi's school announced a Robot Maze Challenge. Robots from different schools would come and try to solve a maze. The robot that solved the maze fastest would win a prize.
Ngozi was excited. "Musa, we are going to win!" she said.
But there was a problem. Musa could avoid obstacles, but he could not plan a path. He would move forward. If he hit a wall, he would turn. If he hit another wall, he would turn again. He would wander around the maze without a plan.
Ngozi watched in frustration. "Musa, you are going in circles!" she said.
Musa did not answer. He just kept wandering.
Ngozi went to her teacher. "Madam, Musa cannot solve the maze. He just wanders around. What can I do?"
Her teacher smiled. "Ngozi, you need to teach Musa about motion planning. You need to teach him to plan a path before he moves."
"What is motion planning?" Ngozi asked.
"Motion planning is the art of finding a path from one place to another without hitting obstacles," her teacher explained. "You need to teach Musa to look at the maze, find a path, and then follow it."
Ngozi was fascinated. She learned about configuration space β a way to describe all possible positions of the robot. She learned about free space β the space where the robot can move. She learned about obstacle space β the space where the robot cannot move.
She learned about roadmap methods β ways to find paths through free space. She learned about RRT β a method that grows a tree of possible paths. She learned about trajectory optimization β a way to make paths smooth.
She wrote a new program for Musa. The program planned a path from the start to the goal. It avoided obstacles. It followed the path smoothly.
She tested the new program. Musa moved forward. He turned left. He went forward again. This time, he did not wander. He followed the planned path. He reached the goal.
On the day of the challenge, Musa performed beautifully. He solved the maze in record time. Ngozi's school won first place.
But more importantly, Ngozi learned something powerful: robots can solve complex problems if you teach them to plan.
That is what you will learn in this module. You will learn how robots plan paths. You will learn about configuration space, roadmap methods, RRT, and trajectory optimization.
Let us begin!
Motion planning is the process of finding a path from a start position to a goal position without hitting obstacles.
Without motion planning, a robot cannot move intelligently. It might bump into things. It might get stuck. It might take a very long path. Motion planning makes robots efficient and safe.
Imagine you want to walk from your house to the market. There are many routes. Some are short. Some are long. Some have obstacles. You choose the best route. That is motion planning.
A self-driving car plans a route from your house to your destination. It avoids traffic and obstacles.
A robot in a science fair plans a path through a maze.
A robot vacuum plans a path to clean your room efficiently.
A danfo driver plans a route to avoid traffic in Lagos.
Motion Planning Start | | (obstacles) V Goal Plan a path from Start to Goal, avoiding obstacles.
Motion planning is the process of finding a path from a start position to a goal position without hitting obstacles.
Configuration space is a way to describe all possible positions of a robot. Each point in configuration space represents one position of the robot.
Configuration space helps us plan robot movements. We can see which positions are free and which are blocked. We can find a path from one position to another.
Imagine a robot arm with one joint. The joint can rotate from 0 degrees to 180 degrees. The configuration space is all the angles between 0 and 180. Each angle is one configuration.
When you open a door, the door can be at many angles. Each angle is a configuration.
A pencil can point in many directions. Each direction is a configuration.
A fan can rotate to many angles. Each angle is a configuration.
A traffic warden can point in many directions. Each direction is a configuration.
Configuration Space
For a 1-joint robot:
0 degrees
|
V
45 degrees
|
V
90 degrees
|
V
135 degrees
|
V
180 degrees
Each angle is a configuration.
Configuration space describes all possible positions of a robot. Each point is one configuration.
Free space is the space where the robot can move. Obstacle space is the space where the robot cannot move because there are obstacles.
Motion planning is about finding a path through free space. If we know where the obstacles are, we can avoid them.
Imagine a room with furniture. The floor space where you can walk is free space. The space taken by furniture is obstacle space.
On a road, the driving lane is free space. The pavement is obstacle space.
In a classroom, the aisles are free space. The desks are obstacle space.
In a kitchen, the floor is free space. The cabinets are obstacle space.
In a market, the walkways are free space. The stalls are obstacle space.
Free Space and Obstacle Space +-------------------+ | O O O O O O O O | | O . . . . . . O | | O . . . . . . O | | O . . . . . . O | | O O O O O O O O | +-------------------+ O = obstacle space . = free space
Free space is where the robot can move. Obstacle space is where the robot cannot move.
A roadmap is a network of paths through free space. Roadmap methods build a graph of possible paths and search for a route from start to goal.
Roadmap methods are simple and effective. They work well for robots that move in 2D or 3D spaces.
Imagine a city map. Roads connect different places. You can travel from one place to another by following roads. A roadmap in motion planning is like a city map for robots.
| Method | How It Works | Example |
|---|---|---|
| Visibility Graph | Connects vertices of obstacles with straight lines | Polygon obstacles |
| Voronoi Diagram | Creates paths that are as far as possible from obstacles | Safety-critical paths |
| Probabilistic Roadmap | Randomly samples points and connects them | High-dimensional spaces |
A GPS uses a roadmap to find a route from your house to your destination.
A robot in a maze uses a roadmap to find its way out.
A robot vacuum uses a roadmap to clean all rooms efficiently.
A delivery robot in Lagos uses a roadmap to navigate streets.
Roadmap Method
Start ----+----+----+
| | |
+----+----+
| | |
+----+----+---- Goal
| | |
+----+----+
Find a path from Start to Goal.
A roadmap is a network of paths through free space. Roadmap methods build a graph of possible paths and search for a route.
A visibility graph is a roadmap that connects the vertices of obstacles with straight lines. Two points are connected if they can "see" each other without hitting an obstacle.
Visibility graphs find the shortest path in a 2D environment with polygon obstacles. They are simple and effective.
Imagine you are in a room with tables. You want to walk from one corner to another. You can see some tables clearly. You cannot see others because they are blocked. You connect the points you can see. That is a visibility graph.
In a park, you can see some benches clearly. You walk in a straight line to them. That is a visibility graph.
In a classroom, you can see the board from your seat. You walk in a straight line to it.
In a living room, you can see the TV from the sofa. You walk in a straight line to it.
In a market, you can see some stalls clearly. You walk in a straight line to them.
Visibility Graph Start | +----+----+ | | | +----+----+ | | | +----+----+---- Goal Connect points that can "see" each other.
A visibility graph connects the vertices of obstacles with straight lines. It finds the shortest path in a 2D environment with polygon obstacles.
A Voronoi diagram is a roadmap that creates paths that are as far as possible from obstacles. It maximizes safety.
Voronoi diagrams are good for safety-critical paths. The robot stays as far as possible from obstacles. This reduces the risk of collisions.
Imagine you are walking in a room with pillars. You want to stay as far as possible from each pillar. You walk in the middle of the free space. That is a Voronoi diagram.
In a museum, you walk in the middle of the corridor to avoid touching the exhibits. That is a Voronoi diagram.
In a classroom, you walk in the middle of the aisle to avoid bumping desks.
In a kitchen, you walk in the middle of the floor to avoid cabinets.
In a market, you walk in the middle of the walkway to avoid stalls.
Voronoi Diagram +-------------------+ | O O O | | \ | / | | \ | / | | \ | / | | * | | / | \ | | / | \ | | / | \ | | O O O | +-------------------+ Paths are as far as possible from obstacles.
A Voronoi diagram creates paths that are as far as possible from obstacles. It maximizes safety.
Cell decomposition divides free space into simple cells. The robot plans a path through the cells.
Cell decomposition simplifies motion planning. Instead of dealing with complex shapes, the robot deals with simple cells.
Imagine a floor plan divided into squares. Each square is a cell. You move from one cell to the next. That is cell decomposition.
A chessboard is divided into squares. Each square is a cell.
A school timetable divides the day into periods. Each period is a cell.
A garden divided into plots. Each plot is a cell.
A farm divided into plots. Each plot is a cell.
Cell Decomposition +---+---+---+---+ | | | | | +---+---+---+---+ | | | | | +---+---+---+---+ | | | | | +---+---+---+---+ Each cell is a simple shape.
Cell decomposition divides free space into simple cells. The robot plans a path through the cells.
RRT is a motion planning algorithm that grows a tree of possible paths. It randomly samples points in free space and connects them.
RRT is fast and can handle complex environments. It is widely used in robotics.
Imagine you are exploring a new forest. You start at one point. You randomly pick a direction. You walk a little. You mark the spot. You repeat. That is RRT.
A search and rescue robot explores a collapsed building using RRT.
A robot in a maze uses RRT to find a path.
A robot vacuum uses RRT to explore a new room.
A drone delivering medicine in a village uses RRT to avoid trees.
RRT
Start
*
\
*
\
*
\
*
\
Goal
Grow a tree by randomly sampling.
RRT grows a tree of possible paths by randomly sampling points in free space.
RRT* is an improved version of RRT. It not only finds a path but also improves it over time. It finds the optimal path.
RRT* finds better paths than RRT. It is used when path quality matters.
Imagine you are exploring a forest. You find a path to the goal. But then you find a shorter path. You update your path. That is RRT*.
A self-driving car uses RRT* to find the shortest route.
A robot in a maze uses RRT* to find the shortest path.
A robot vacuum uses RRT* to clean efficiently.
A delivery robot in Lagos uses RRT* to find the fastest route.
RRT*
Start
*
\
*----*
\ \
*----*---- Goal
\
*
Improves the path over time.
RRT* is an improved version of RRT. It finds the optimal path by improving it over time.
Trajectory optimization is the process of making a path smooth and efficient. It considers the robot's dynamics.
Raw paths from RRT might be jerky. Trajectory optimization makes them smooth. The robot moves faster and uses less energy.
Imagine you are drawing a line. First you draw it roughly. Then you smooth it out. That is trajectory optimization.
A car's cruise control smooths the ride.
A robot arm smooths its motion to avoid vibration.
A robot vacuum smooths its motion to clean quietly.
A drone smooths its flight to avoid crashing.
Trajectory Optimization Raw path: * * * * * \/ \/ \/ \/ Smoothed path: *----------* Make the path smooth.
Trajectory optimization makes a path smooth and efficient. It considers the robot's dynamics.
Path tracking is the process of following a planned path. The robot uses controllers to stay on the path.
Even with a good plan, the robot might drift off the path. Path tracking keeps the robot on track.
Imagine you are driving a car. You have a route. But you might drift. You steer to stay on the road. That is path tracking.
A self-driving car uses path tracking to stay in its lane.
A robot in a maze uses path tracking to follow the planned path.
A robot vacuum uses path tracking to clean in straight lines.
A danfo driver uses path tracking to stay on the road.
Path Tracking
Planned path:
*----------*
Actual path:
*--\ /--*
\/
Controller keeps robot on path.
Path tracking is the process of following a planned path. Controllers keep the robot on track.
Model Predictive Control is a control method that predicts the future and plans accordingly. It is used for trajectory tracking.
MPC is powerful. It handles constraints and optimizes performance. It is used in self-driving cars and drones.
Imagine you are driving. You look ahead. You predict what will happen. You adjust your speed and direction. That is MPC.
A self-driving car uses MPC to plan its next moves.
A robot in a maze uses MPC to plan ahead.
A robot vacuum uses MPC to clean efficiently.
A delivery robot in Lagos uses MPC to avoid traffic.
Model Predictive Control
Current state
|
V
Predict future
|
V
Optimize plan
|
V
Apply control
|
V
Repeat
Model Predictive Control predicts the future and plans accordingly. It is used for trajectory tracking.
Real robots use motion planning to move safely and efficiently.
| Robot | Motion Planning Method | Purpose |
|---|---|---|
| Robot Vacuum | RRT, SLAM | Clean floors |
| Self-Driving Car | RRT*, MPC | Drive on roads |
| Mars Rover | Visibility Graph, RRT | Explore Mars |
| Surgical Robot | Trajectory Optimization | Perform surgery |
| Drone | RRT*, MPC | Fly and deliver |
A robot vacuum uses SLAM and RRT to clean your house.
A school robot uses RRT* to solve a maze.
A robot toy uses RRT to follow you.
A delivery robot in Lagos uses MPC to avoid traffic.
Real Robot: Self-Driving Car
+-------------------+
| Sensors |
| (see world) |
+-------------------+
|
V
+-------------------+
| Motion Planner |
| (RRT*, MPC) |
+-------------------+
|
V
+-------------------+
| Control |
+-------------------+
|
V
+-------------------+
| Car moves |
+-------------------+
Real robots use motion planning to move safely and efficiently. Different methods are used for different tasks.
Debugging motion planning problems means finding and fixing issues in path planning.
Motion planning can fail. The robot might not find a path. The path might be too long. The robot might hit an obstacle. Debugging helps you fix these problems.
| Problem | Cause | Solution |
|---|---|---|
| No path found | Goal is in obstacle space | Check goal position |
| Path is too long | Poor sampling | Increase samples or use RRT* |
| Robot hits obstacle | Path not collision-free | Add collision checking |
| Robot moves jerky | No trajectory optimization | Add smoothing |
| Robot drifts off path | Poor path tracking | Tune controller |
If your GPS cannot find a route, check the destination address.
If your robot cannot find a path in a maze, check the goal position.
If your robot vacuum gets stuck, check the map.
If a delivery robot in Lagos cannot find a route, check the road conditions.
Debugging Motion Planning
[ No path found ]
|
V
[ Check goal position ]
|
V
[ Check obstacles ]
|
V
[ Check planner settings ]
|
V
[ Fix problem ]
|
V
[ Test again ]
Debugging motion planning problems means finding and fixing issues in path planning. Common problems include no path found, long paths, and collisions.
The future of motion planning is very exciting. Robots will become smarter and more capable.
Better motion planning means better robots. They will work in more complex environments. They will help people in more ways.
Self-driving cars are already using advanced motion planning.
Students today are learning motion planning for future jobs.
Robot vacuums are becoming smarter every year.
Nigerian universities are researching motion planning for agriculture and healthcare.
Future of Motion Planning
Today: Simple paths
|
V
Soon: Complex paths
|
V
Future: Real-time planning
|
V
Future: Multi-robot planning
The future of motion planning is exciting. Robots will become smarter and more capable.
| Word | Simple Definition |
|---|---|
| Motion Planning | Finding a path from start to goal without hitting obstacles. |
| Configuration Space | All possible positions of a robot. |
| Free Space | Space where the robot can move. |
| Obstacle Space | Space where the robot cannot move. |
| Roadmap | A network of paths through free space. |
| Visibility Graph | Connects vertices of obstacles with straight lines. |
| Voronoi Diagram | Creates paths as far as possible from obstacles. |
| Cell Decomposition | Divides free space into simple cells. |
| RRT | Rapidly-exploring Random Tree. |
| RRT* | Improved version of RRT that finds optimal paths. |
| Trajectory Optimization | Makes a path smooth and efficient. |
| Path Tracking | Following a planned path. |
| MPC | Model Predictive Control. |
| Concept | Real-Life Example |
|---|---|
| Motion Planning | Self-driving car plans a route. |
| Configuration Space | A door can be at many angles. |
| Free Space | The driving lane on a road. |
| Obstacle Space | The pavement on a road. |
| Roadmap | GPS uses a roadmap. |
| Visibility Graph | Walking in a straight line to a visible point. |
| Voronoi Diagram | Walking in the middle of a corridor. |
| Cell Decomposition | A chessboard divided into squares. |
| RRT | Search and rescue robot explores a building. |
| RRT* | Self-driving car finds shortest route. |
| Trajectory Optimization | Car's cruise control smooths the ride. |
| Path Tracking | Self-driving car stays in lane. |
| Concept | Nigerian Example |
|---|---|
| Motion Planning | A danfo driver plans a route to avoid traffic. |
| Configuration Space | A traffic warden can point in many directions. |
| Free Space | The walkway in a market. |
| Obstacle Space | The stalls in a market. |
| Roadmap | A delivery robot uses a roadmap. |
| Visibility Graph | Walking in a straight line to a visible stall. |
| Voronoi Diagram | Walking in the middle of a market walkway. |
| Cell Decomposition | A farm divided into plots. |
| RRT | A drone avoids trees in a village. |
| RRT* | A delivery robot finds fastest route. |
| Trajectory Optimization | A drone smooths its flight. |
| Path Tracking | A danfo driver stays on the road. |
| Concept | Everyday Example |
|---|---|
| Motion Planning | Planning your route to school. |
| Configuration Space | A fan can rotate to many angles. |
| Free Space | The floor in your room. |
| Obstacle Space | The furniture in your room. |
| Roadmap | A map of your neighbourhood. |
| Visibility Graph | Walking in a straight line to a visible point. |
| Voronoi Diagram | Walking in the middle of a corridor. |
| Cell Decomposition | A garden divided into plots. |
| RRT | Exploring a new neighbourhood. |
| RRT* | Finding a shortcut to school. |
| Trajectory Optimization | Smoothing your walk. |
| Path Tracking | Staying on the sidewalk. |
| Mistake | Why It Is Wrong | How to Fix It |
|---|---|---|
| Not checking for collisions | Robot hits obstacles. | Add collision checking. |
| Using too few samples | Path is poor. | Increase samples. |
| Ignoring dynamics | Path is not feasible. | Consider robot dynamics. |
| Not smoothing path | Robot moves jerky. | Add trajectory optimization. |
| Poor path tracking | Robot drifts off path. | Tune controller. |
| Not debugging | Problems continue. | Find and fix problems. |
+-------------------+
| Start |
+-------------------+
|
V
+-------------------+
| Motion Planner |
| (finds path) |
+-------------------+
|
V
+-------------------+
| Path |
+-------------------+
|
V
+-------------------+
| Trajectory |
| Optimization |
+-------------------+
|
V
+-------------------+
| Path Tracking |
+-------------------+
|
V
+-------------------+
| Goal |
+-------------------+
( Start )
|
V
+----------------+
| Initialize |
| tree with root |
+----------------+
|
V
+----------------+
| Sample random |
| point |
+----------------+
|
V
+----------------+
| Find nearest |
| node |
+----------------+
|
V
+----------------+
| Extend tree |
+----------------+
|
V
+----------------+
| Collision? |
+----------------+
/ \
YES NO
/ \
V V
+---------+ +-----------+
| Discard | | Add node |
+---------+ +-----------+
\ /
\ /
V V
+----------------+
| Goal reached? |
+----------------+
/ \
YES NO
/ \
V V
+---------+ +-----------+
| Extract | | Repeat |
| path | | |
+---------+ +-----------+
| Method | Speed | Optimality | Best For |
|---|---|---|---|
| Visibility Graph | Fast | Optimal | 2D polygon obstacles |
| Voronoi Diagram | Fast | Safe | Safety-critical paths |
| Cell Decomposition | Medium | Good | Simple environments |
| RRT | Fast | Not optimal | Complex environments |
| RRT* | Slower | Optimal | When path quality matters |
| Trajectory Optimization | Slow | Optimal | Smooth paths |
Step 1: Define start and goal
|
V
Step 2: Build configuration space
|
V
Step 3: Identify free space
|
V
Step 4: Choose planning method
|
V
Step 5: Find path
|
V
Step 6: Optimize path
|
V
Step 7: Track path
|
V
Step 8: Reach goal
Motion planning is the process of finding a path from a start position to a goal position without hitting obstacles.
Configuration space describes all possible positions of a robot.
Free space is where the robot can move. Obstacle space is where the robot cannot move.
A roadmap is a network of paths through free space. Roadmap methods build a graph of possible paths and search for a route.
A visibility graph connects the vertices of obstacles with straight lines. It finds the shortest path.
A Voronoi diagram creates paths that are as far as possible from obstacles. It maximizes safety.
Cell decomposition divides free space into simple cells. The robot plans a path through the cells.
RRT grows a tree of possible paths by randomly sampling points in free space.
RRT* is an improved version of RRT. It finds the optimal path by improving it over time.
Trajectory optimization makes a path smooth and efficient. It considers the robot's dynamics.
Path tracking is the process of following a planned path. Controllers keep the robot on track.
Model Predictive Control predicts the future and plans accordingly. It is used for trajectory tracking.
Real robots use motion planning to move safely and efficiently.
Debugging motion planning problems means finding and fixing issues in path planning.
The future of motion planning is exciting. Robots will become smarter and more capable.
In this module, you learned about motion planning. You learned that motion planning is the process of finding a path from a start position to a goal position without hitting obstacles.
You learned about configuration space β a way to describe all possible positions. You learned about free space and obstacle space. You learned about roadmap methods, visibility graphs, and Voronoi diagrams. You learned about cell decomposition.
You learned about RRT β a fast method that grows a tree of possible paths. You learned about RRT* β an improved version that finds optimal paths. You learned about trajectory optimization β making paths smooth and efficient. You learned about path tracking and Model Predictive Control.
You learned about real robots and their motion planning. You learned about debugging motion planning problems. You learned about the future of motion planning.
Most importantly, you learned that motion planning is what allows robots to move intelligently in complex environments.
In the next module, you will learn about Robot Learning. You will learn how robots learn from experience. You will learn about reinforcement learning, imitation learning, and deep learning.
But for now, take a moment to celebrate what you have learned. You have taken another big step in your journey to becoming a robotics expert. Well done!
Match the term on the left with its definition on the right.
| Term | Definition |
|---|---|
| 1. Motion Planning | A. All possible positions of a robot |
| 2. Configuration Space | B. Finding a path from start to goal |
| 3. Free Space | C. Space where the robot cannot move |
| 4. Obstacle Space | D. Space where the robot can move |
| 5. Roadmap | E. Grows a tree of possible paths |
| 6. RRT | F. A network of paths through free space |
| 7. RRT* | G. Makes a path smooth and efficient |
| 8. Trajectory Optimization | H. Improved version of RRT |
Answers: 1-B, 2-A, 3-D, 4-C, 5-F, 6-E, 7-H, 8-G
Title: Motion Planning Game
Instructions:
Example:
Maze: +---+---+---+---+ | S | | O | | +---+---+---+---+ | | O | | O | +---+---+---+---+ | O | | | | +---+---+---+---+ | | | O | G | +---+---+---+---+ S = Start G = Goal O = Obstacle Plan a path from S to G.
Title: Motion Planning Scavenger Hunt
Instructions:
Example:
| Situation | Obstacles | Path |
|---|---|---|
| Walking to school | Cars, potholes | Sidewalk |
| Cleaning a room | Furniture | Around furniture |
| Driving to market | Traffic, pedestrians | Main road |
| Delivering a package | Buildings, trees | Streets |
| Flying a drone | Trees, buildings | Open air |
Title: Build a Simple Motion Planner
Goal: Create a program that plans a path from start to goal in a simple 2D environment.
Steps:
Deliverables:
Title: Implement RRT* in Simulation
Instructions:
Grading Criteria:
| Criteria | Points |
|---|---|
| RRT* works correctly | 30 |
| Finds optimal paths | 25 |
| Comparison with RRT | 15 |
| Report is clear | 15 |
| Visualization is clear | 15 |
| Total | 100 |
In Module Four, you will learn about Robot Learning. You will learn how robots learn from experience.
You will learn about:
To prepare for Module Four:
Get ready for an exciting journey into the world of robot learning!
Congratulations! You have completed Module Three of Fundamentals of Robotics Level Three. You have learned about motion planning.
You learned that motion planning is the process of finding a path from a start position to a goal position without hitting obstacles. You learned about configuration space, free space, and obstacle space.
You learned about roadmap methods, visibility graphs, and Voronoi diagrams. You learned about cell decomposition. You learned about RRT and RRT*. You learned about trajectory optimization, path tracking, and Model Predictive Control.
You learned about real robots and their motion planning. You learned about debugging motion planning problems. You learned about the future of motion planning.
You also learned many examples from Nigeria, from your home, from school, and from everyday life. You learned through stories, illustrations, and activities.
Now you are ready for Module Four: Robot Learning. In Module Four, you will learn how robots learn from experience. You will learn about reinforcement learning, imitation learning, and deep learning.
But before you move on, take a moment to review this module. Make sure you understand the key ideas. Practise motion planning problems. Draw diagrams. Test your understanding. The more you practise, the better you will become.
You are doing great. Keep learning. Keep exploring. Keep building. The world of robotics is waiting for you!
End of Module Three
Next: Module Four β Robot Learning
Welcome to Module Four of Level Three! In Module One, you learned about kinematics and dynamics. In Module Two, you learned about probabilistic robotics. In Module Three, you learned about motion planning. Now you will learn about robot learning.
Robot learning is the study of how robots can learn from experience. Instead of being programmed for every situation, a robot can learn by trying, failing, and improving. It is like how you learn to ride a bicycle. You do not read a manual. You try, you fall, you try again, and eventually you succeed.
Think about how you learned to walk. When you were a baby, you did not know how. You tried. You fell. You tried again. Slowly, you learned. Robots can learn the same way.
Robot learning is one of the most exciting areas of robotics. It is used in self-driving cars, robot arms, drones, and even robots that play games. In this module, you will learn about reinforcement learning, Markov Decision Processes, Q-Learning, deep reinforcement learning, and imitation learning.
Do not worry if these words sound hard. We will explain everything step by step, using simple examples and stories.
Let us begin!
By the end of this module, you will be able to:
Once upon a time, in the city of Port Harcourt, Nigeria, there lived a boy named Tunde. Tunde had built a small robot named Kemi. Kemi had two legs, motors, sensors, and a small computer brain.
Tunde wanted Kemi to walk. But walking is hard. You have to balance. You have to move your legs. You have to avoid falling. Tunde did not know how to program all of this.
He tried writing a program. The program said: "Move left leg forward. Move right leg forward. Repeat." But Kemi fell over. She could not balance.
Tunde tried again. He added more instructions. But Kemi still fell. The program was too simple. Walking is too complex to program step by step.
Tunde went to his teacher. "Madam, I cannot program Kemi to walk. It is too hard."
His teacher smiled. "Tunde, you do not need to program every step. You need to let Kemi learn."
"What do you mean?" Tunde asked.
"Use reinforcement learning," his teacher said. "Give Kemi a reward when she walks. Give her a penalty when she falls. She will learn to walk by trying."
Tunde was fascinated. He learned about reinforcement learning. He learned about rewards and penalties. He learned about Markov Decision Processes. He learned about Q-Learning.
He wrote a new program. The program did not tell Kemi exactly what to do. Instead, it gave Kemi a reward when she moved forward. It gave her a penalty when she fell.
At first, Kemi fell a lot. She did not know what to do. But slowly, she learned. She learned to balance. She learned to move her legs. She learned to walk.
After many tries, Kemi took her first steps. She walked across the room. She did not fall. Tunde cheered.
"Kemi, you learned to walk!" he shouted.
Kemi kept walking. She was no longer programmed. She had learned.
That is what you will learn in this module. You will learn how robots learn from experience. You will learn about reinforcement learning, Q-Learning, deep reinforcement learning, and imitation learning.
Let us begin!
Robot learning is the process of a robot improving its behaviour through experience. Instead of being programmed for every situation, the robot learns from its actions and their results.
Some tasks are too complex to program step by step. Walking, grasping, and driving are examples. Robot learning allows robots to figure out these tasks on their own.
Think about learning to ride a bicycle. Nobody can tell you exactly how to balance. You have to try. You fall. You try again. Slowly, you learn. Robots learn the same way.
A self-driving car learns to drive by practising in simulation.
A student learns to solve math problems by practising.
You learn to cook by trying recipes and improving.
A trader learns which goods sell best by trying different products.
Robot Learning Try --> Fail --> Learn --> Try Again --> Succeed The robot improves with each attempt.
Robot learning is the process of a robot improving its behaviour through experience.
Programming means telling the robot exactly what to do. Learning means the robot figures out what to do by itself.
Programming is good for simple tasks. Learning is good for complex tasks. Knowing the difference helps you choose the right approach.
| Aspect | Programming | Learning |
|---|---|---|
| Who decides? | Human programmer | Robot itself |
| Best for | Simple, clear tasks | Complex, unclear tasks |
| Example | Blink a light | Walk on two legs |
| Flexibility | Low | High |
Programming is like following a recipe. Learning is like experimenting in the kitchen.
A calculator is programmed. A chess computer learns.
Memorising times tables is programming. Solving new problems is learning.
Following a recipe is programming. Inventing a new dish is learning.
Following a bus route is programming. Finding a shortcut is learning.
Programming vs Learning Programming: Human --> Instructions --> Robot Learning: Robot --> Try --> Feedback --> Improve
Programming means telling the robot exactly what to do. Learning means the robot figures out what to do by itself.
Reinforcement learning is a type of learning where the robot learns by receiving rewards and penalties for its actions.
Reinforcement learning is powerful. It allows robots to learn complex tasks without being told exactly what to do.
Think of training a dog. When the dog sits, you give it a treat. When it does something wrong, you say "No." The dog learns to sit because it wants the treat. Robots learn the same way.
| Term | Meaning |
|---|---|
| Agent | The robot or learner |
| Environment | The world the robot interacts with |
| Action | What the robot does |
| Reward | Positive feedback for good action |
| Penalty | Negative feedback for bad action |
| State | The current situation |
A robot learns to walk by getting a reward for moving forward and a penalty for falling.
You learn to study by getting good grades (reward) and avoiding bad grades (penalty).
You learn to cook by getting compliments (reward) and avoiding burnt food (penalty).
A trader learns which goods to sell by making profit (reward) and avoiding losses (penalty).
Reinforcement Learning
Agent (Robot)
|
| Action
V
Environment
|
| Reward/Penalty
V
Agent learns
|
| Repeat
V
Reinforcement learning is a type of learning where the robot learns by receiving rewards and penalties for its actions.
A Markov Decision Process (MDP) is a mathematical framework for reinforcement learning. It describes the states, actions, and rewards in a learning problem.
MDPs help us model learning problems. They tell us what the robot can do, what it knows, and what it wants.
Imagine a board game. You have a position (state). You can move (action). You get points (reward). The rules of the game are the MDP.
A chess game is an MDP. States are board positions. Actions are moves. Rewards are winning or losing.
Studying for an exam is an MDP. States are knowledge levels. Actions are study choices. Rewards are grades.
Cooking is an MDP. States are cooking stages. Actions are cooking steps. Rewards are taste.
A trader's day is an MDP. States are stock levels. Actions are buying and selling. Rewards are profit.
Markov Decision Process
State 1 --> Action --> State 2
|
V
Reward
State 2 --> Action --> State 3
|
V
Reward
A Markov Decision Process is a mathematical framework for reinforcement learning. It describes the states, actions, and rewards in a learning problem.
Rewards are positive feedback. Penalties are negative feedback. Both help the robot learn.
Rewards and penalties tell the robot what is good and what is bad. Without them, the robot cannot learn.
Think of a video game. When you collect a coin, you get points (reward). When you hit an enemy, you lose points (penalty). The game teaches you what to do.
A robot gets a reward for reaching a goal and a penalty for hitting a wall.
You get a reward for correct answers and a penalty for wrong ones.
You get a reward for cleaning your room and a penalty for not doing chores.
A trader gets a reward for profit and a penalty for loss.
Rewards and Penalties Action: Move forward Result: Reached goal Reward: +10 Action: Hit wall Result: Robot stuck Penalty: -5 Robot learns to move forward, not hit walls.
Rewards are positive feedback. Penalties are negative feedback. Both help the robot learn.
Q-Learning is a reinforcement learning algorithm. It learns the value of taking an action in a state. This value is called the Q-value.
Q-Learning is simple and effective. It is used in many robot learning applications.
Imagine you are in a maze. You want to find the exit. You try different paths. You remember which paths lead to the exit. Q-Learning helps you remember.
A robot learns the best path to a goal using Q-Learning.
You learn the best way to study using Q-Learning principles.
You learn the best way to organise your room.
A trader learns the best market to sell in.
Q-Learning State: At start Actions: Left, Right, Forward Q-values: Left: 0.5 Right: 0.2 Forward: 0.8 <-- Best Choose Forward.
Q-Learning is a reinforcement learning algorithm that learns the value of taking an action in a state.
SARSA is another reinforcement learning algorithm. It stands for State-Action-Reward-State-Action. It learns by looking at the action the robot actually takes.
SARSA is safer than Q-Learning. It considers the robot's actual behaviour, not just the best possible behaviour.
Imagine you are following a path. Q-Learning asks: "What is the best path?" SARSA asks: "What is the best path given what I actually do?"
A robot learns to walk carefully using SARSA.
You learn to study consistently using SARSA.
You learn to save money carefully.
A trader learns to manage stock carefully.
SARSA State --> Action --> Reward --> State --> Action S --> A --> R --> S' --> A' Update based on actual next action.
SARSA is a reinforcement learning algorithm that learns by looking at the action the robot actually takes.
Deep reinforcement learning combines reinforcement learning with deep neural networks. It allows robots to learn from complex inputs like images.
Deep reinforcement learning is powerful. It can handle complex tasks like driving and playing games.
Imagine you are learning to play a video game. You see the screen (images). You decide what to do. Deep reinforcement learning works the same way.
A self-driving car uses deep reinforcement learning to drive.
A robot learns to play chess using deep reinforcement learning.
A robot vacuum learns to clean efficiently.
A drone learns to deliver packages in a city.
Deep Reinforcement Learning
Images --> Neural Network --> Actions
|
V
Rewards
|
V
Learning
Deep reinforcement learning combines reinforcement learning with deep neural networks. It allows robots to learn from complex inputs like images.
Imitation learning is a type of learning where the robot learns by watching a human or another robot. It copies the behaviour it sees.
Imitation learning is fast. The robot does not have to try everything from scratch. It can learn from an expert.
Think about how you learned to write. You watched your teacher write. You copied. That is imitation learning.
A robot learns to cook by watching a chef.
You learn to solve math problems by watching your teacher.
You learn to cook by watching your mother.
An apprentice learns a trade by watching a master.
Imitation Learning Human demonstrates --> Robot observes --> Robot copies The robot learns by watching.
Imitation learning is a type of learning where the robot learns by watching a human or another robot.
Sim-to-real transfer means training a robot in simulation, then using what it learned in the real world.
Training in the real world is slow and expensive. Training in simulation is fast and cheap. Sim-to-real transfer saves time and money.
Think of learning to drive in a video game. Then you drive a real car. That is sim-to-real transfer.
A drone learns to fly in simulation, then flies in the real world.
You practise a sport in a video game, then play the real sport.
You practise cooking in a game, then cook real food.
A pilot trains on a flight simulator, then flies a real plane.
Sim-to-Real Transfer Simulation --> Learn --> Real World Train in simulation, apply in reality.
Sim-to-real transfer means training a robot in simulation, then using what it learned in the real world.
Transfer learning means using what you learned in one task to help with another task.
Transfer learning saves time. The robot does not have to start from scratch for every new task.
If you know how to ride a bicycle, you can learn to ride a motorcycle faster. That is transfer learning.
A robot that learned to pick up balls can learn to pick up cups faster.
If you know algebra, you can learn calculus faster.
If you know how to cook rice, you can learn to cook beans faster.
If you know how to sew dresses, you can learn to sew shirts faster.
Transfer Learning
Task 1: Learn to pick balls
|
V
Task 2: Learn to pick cups (faster!)
Knowledge transfers from Task 1 to Task 2.
Transfer learning means using what you learned in one task to help with another task.
Active learning means the robot chooses which examples to learn from. It asks for help when it is unsure.
Active learning is efficient. The robot focuses on the examples that matter most.
Imagine you are studying for a test. Instead of reading the whole book, you focus on the parts you do not understand. That is active learning.
A robot asks a human for help when it encounters a new object.
You ask your teacher for help when you are stuck.
You ask your mother for help when you are cooking a new dish.
A trader asks customers what they want.
Active Learning Robot is unsure --> Asks for help --> Learns The robot chooses what to learn.
Active learning means the robot chooses which examples to learn from. It asks for help when it is unsure.
Real robots use learning to improve their behaviour.
| Robot | Learning Method | Purpose |
|---|---|---|
| Self-Driving Car | Deep reinforcement learning | Drive safely |
| Robot Arm | Imitation learning | Pick and place |
| Robot Vacuum | Reinforcement learning | Clean efficiently |
| Drone | Sim-to-real transfer | Fly and deliver |
| Game-Playing Robot | Deep reinforcement learning | Play games |
A robot arm learns to pick up objects by watching a human.
A school robot learns to navigate a maze using reinforcement learning.
A robot vacuum learns the layout of your house.
A delivery robot in Lagos learns to avoid traffic.
Real Robot: Self-Driving Car
+-------------------+
| Sensors |
| (see world) |
+-------------------+
|
V
+-------------------+
| Deep RL |
| (learns to |
| drive) |
+-------------------+
|
V
+-------------------+
| Control |
+-------------------+
|
V
+-------------------+
| Car moves |
+-------------------+
Real robots use learning to improve their behaviour. Different methods are used for different tasks.
Debugging robot learning systems means finding and fixing problems in learning algorithms.
Learning systems can fail. The robot might not learn. It might learn the wrong thing. Debugging helps you fix these problems.
| Problem | Cause | Solution |
|---|---|---|
| Robot does not learn | Rewards too small | Increase rewards |
| Robot learns wrong behaviour | Reward function wrong | Fix reward function |
| Robot forgets old skills | Catastrophic forgetting | Use transfer learning |
| Learning is too slow | Too many states | Simplify state space |
| Robot overfits | Too little training data | Add more data |
If a robot does not learn to walk, check the reward function.
If a student does not improve, check the study method.
If a recipe does not work, check the ingredients.
If a trader is not making profit, check the goods and prices.
Debugging Robot Learning
[ Robot not learning ]
|
V
[ Check rewards ]
|
V
[ Check state space ]
|
V
[ Check algorithm ]
|
V
[ Fix problem ]
|
V
[ Test again ]
Debugging robot learning systems means finding and fixing problems in learning algorithms. Common problems include no learning, wrong behaviour, and slow learning.
The future of robot learning is very exciting. Robots will become smarter and more capable.
Better learning means better robots. They will work in more complex environments. They will help people in more ways.
Researchers are already teaching robots to learn from YouTube videos.
Students today are learning robot learning for future jobs.
Robot vacuums are becoming smarter every year.
Nigerian universities are researching robot learning for agriculture and healthcare.
Future of Robot Learning
Today: Simple learning
|
V
Soon: Learning from videos
|
V
Future: Continuous learning
|
V
Future: Creative robots
The future of robot learning is exciting. Robots will become smarter and more capable.
| Word | Simple Definition |
|---|---|
| Robot Learning | Improving behaviour through experience. |
| Reinforcement Learning | Learning from rewards and penalties. |
| Agent | The robot or learner. |
| Environment | The world the robot interacts with. |
| Action | What the robot does. |
| Reward | Positive feedback. |
| Penalty | Negative feedback. |
| State | The current situation. |
| Markov Decision Process | A framework for reinforcement learning. |
| Q-Learning | Learning the value of actions. |
| SARSA | Learning from actual actions. |
| Deep Reinforcement Learning | Reinforcement learning with neural networks. |
| Imitation Learning | Learning by watching others. |
| Sim-to-Real Transfer | Training in simulation, applying in reality. |
| Transfer Learning | Using knowledge from one task for another. |
| Active Learning | Choosing what to learn. |
| Concept | Real-Life Example |
|---|---|
| Robot Learning | Self-driving car learns to drive. |
| Programming vs Learning | Calculator is programmed, chess computer learns. |
| Reinforcement Learning | Training a dog with treats. |
| Markov Decision Process | A chess game. |
| Rewards and Penalties | Video game points. |
| Q-Learning | Robot learns best path in a maze. |
| SARSA | Robot learns to walk carefully. |
| Deep Reinforcement Learning | Self-driving car uses images. |
| Imitation Learning | Robot learns to cook by watching a chef. |
| Sim-to-Real Transfer | Drone learns in simulation, flies in reality. |
| Transfer Learning | Knowing bicycle helps learn motorcycle. |
| Active Learning | Robot asks for help when unsure. |
| Concept | Nigerian Example |
|---|---|
| Robot Learning | A trader learns which goods sell best. |
| Programming vs Learning | Following a bus route vs finding a shortcut. |
| Reinforcement Learning | A trader learns from profit and loss. |
| Markov Decision Process | A trader's day: states, actions, rewards. |
| Rewards and Penalties | Profit is reward, loss is penalty. |
| Q-Learning | Trader learns best market to sell in. |
| SARSA | Trader learns to manage stock carefully. |
| Deep Reinforcement Learning | A drone learns to deliver packages. |
| Imitation Learning | Apprentice learns trade by watching master. |
| Sim-to-Real Transfer | Pilot trains on simulator, flies real plane. |
| Transfer Learning | Sewing dresses helps learn to sew shirts. |
| Active Learning | Trader asks customers what they want. |
| Concept | Everyday Example |
|---|---|
| Robot Learning | Learning to cook by trying recipes. |
| Programming vs Learning | Memorising times tables vs solving new problems. |
| Reinforcement Learning | Getting compliments for good cooking. |
| Markov Decision Process | Studying for an exam. |
| Rewards and Penalties | Rewards for cleaning, penalties for not. |
| Q-Learning | Learning the best way to study. |
| SARSA | Learning to save money carefully. |
| Deep Reinforcement Learning | Robot vacuum learns to clean. |
| Imitation Learning | Learning to cook by watching your mother. |
| Sim-to-Real Transfer | Practising cooking in a game, then cooking real food. |
| Transfer Learning | Knowing how to cook rice helps cook beans. |
| Active Learning | Asking for help when cooking a new dish. |
| Mistake | Why It Is Wrong | How to Fix It |
|---|---|---|
| Rewards too small | Robot does not learn. | Increase rewards. |
| Wrong reward function | Robot learns wrong behaviour. | Fix reward function. |
| Too many states | Learning is too slow. | Simplify state space. |
| Too little data | Robot overfits. | Add more data. |
| Ignoring sim-to-real gap | Robot fails in real world. | Add noise to simulation. |
| Not debugging | Problems continue. | Find and fix problems. |
+-------------------+
| Agent |
| (Robot) |
+-------------------+
|
| Action
V
+-------------------+
| Environment |
+-------------------+
|
| Reward / New State
V
+-------------------+
| Agent updates |
| its knowledge |
+-------------------+
|
| Repeat
V
( Start )
|
V
+----------------+
| Initialize |
| Q-table |
+----------------+
|
V
+----------------+
| Observe state |
+----------------+
|
V
+----------------+
| Choose action |
+----------------+
|
V
+----------------+
| Observe reward |
| and new state |
+----------------+
|
V
+----------------+
| Update Q-value |
+----------------+
|
V
+----------------+
| Goal reached? |
+----------------+
/ \
YES NO
/ \
V V
+---------+ +-----------+
| Done | | Repeat |
+---------+ +-----------+
| Method | How It Learns | Best For |
|---|---|---|
| Reinforcement Learning | Rewards and penalties | Complex tasks |
| Q-Learning | Action values | Navigation |
| SARSA | Actual actions | Safe learning |
| Deep RL | Neural networks | Images, complex inputs |
| Imitation Learning | Watching experts | Fast learning |
| Sim-to-Real | Simulation then reality | Expensive tasks |
| Transfer Learning | Reusing knowledge | Similar tasks |
| Active Learning | Choosing examples | Efficient learning |
Step 1: Define task
|
V
Step 2: Choose learning method
|
V
Step 3: Set up rewards
|
V
Step 4: Train in simulation
|
V
Step 5: Test in reality
|
V
Step 6: Debug
|
V
Step 7: Improve
|
V
Step 8: Deploy
Robot learning is the process of a robot improving its behaviour through experience.
Programming means telling the robot exactly what to do. Learning means the robot figures out what to do by itself.
Reinforcement learning is a type of learning where the robot learns by receiving rewards and penalties.
A Markov Decision Process is a mathematical framework for reinforcement learning.
Rewards are positive feedback. Penalties are negative feedback. Both help the robot learn.
Q-Learning is a reinforcement learning algorithm that learns the value of taking an action in a state.
SARSA is a reinforcement learning algorithm that learns by looking at the action the robot actually takes.
Deep reinforcement learning combines reinforcement learning with deep neural networks.
Imitation learning is a type of learning where the robot learns by watching a human or another robot.
Sim-to-real transfer means training a robot in simulation, then using what it learned in the real world.
Transfer learning means using what you learned in one task to help with another task.
Active learning means the robot chooses which examples to learn from. It asks for help when it is unsure.
Real robots use learning to improve their behaviour.
Debugging robot learning systems means finding and fixing problems in learning algorithms.
The future of robot learning is exciting. Robots will become smarter and more capable.
In this module, you learned about robot learning. You learned that robot learning is the process of a robot improving its behaviour through experience.
You learned the difference between programming and learning. You learned about reinforcement learning β learning from rewards and penalties. You learned about Markov Decision Processes. You learned about Q-Learning and SARSA.
You learned about deep reinforcement learning β using neural networks. You learned about imitation learning β learning by watching. You learned about sim-to-real transfer, transfer learning, and active learning.
You learned about real robots and their learning methods. You learned about debugging robot learning systems. You learned about the future of robot learning.
Most importantly, you learned that robot learning is what allows robots to handle complex tasks that cannot be programmed step by step.
In the next module, you will learn about Manipulation and Grasping. You will learn how robots pick up and move objects. You will learn about contact modeling, grasping, and dexterous manipulation.
But for now, take a moment to celebrate what you have learned. You have taken another big step in your journey to becoming a robotics expert. Well done!
Match the term on the left with its definition on the right.
| Term | Definition |
|---|---|
| 1. Robot Learning | A. Learning from rewards and penalties |
| 2. Reinforcement Learning | B. Improving behaviour through experience |
| 3. Q-Learning | C. Learning by watching others |
| 4. SARSA | D. Training in simulation, applying in reality |
| 5. Imitation Learning | E. Learning the value of actions |
| 6. Sim-to-Real Transfer | F. Learning from actual actions |
| 7. Transfer Learning | G. Choosing what to learn |
| 8. Active Learning | H. Using knowledge from one task for another |
Answers: 1-B, 2-A, 3-E, 4-F, 5-C, 6-D, 7-H, 8-G
Title: Design a Reward System
Instructions:
Example:
Task: Robot cleans a room Rewards: +10 for picking up an object +5 for moving to a new area -5 for hitting furniture -10 for falling over Goal: Maximize total reward.
Title: Learning Scavenger Hunt
Instructions:
Example:
| Thing | What It Learns | How It Learns |
|---|---|---|
| You | New skills | Practice |
| Robot vacuum | House layout | SLAM |
| Phone | Your habits | Data analysis |
| Game AI | Player behaviour | Reinforcement learning |
| Weather app | Weather patterns | Data analysis |
Title: Build a Simple Q-Learning Agent
Goal: Create a program that learns to solve a simple maze using Q-Learning.
Steps:
Deliverables:
Title: Implement Reinforcement Learning in Simulation
Instructions:
Grading Criteria:
| Criteria | Points |
|---|---|
| Learning algorithm works | 30 |
| Robot learns the task | 25 |
| Reward function is well designed | 15 |
| Report is clear | 15 |
| Visualization is clear | 15 |
| Total | 100 |
In Module Five, you will learn about Manipulation and Grasping. You will learn how robots pick up and move objects.
You will learn about:
To prepare for Module Five:
Get ready for an exciting journey into the world of robot manipulation!
Congratulations! You have completed Module Four of Fundamentals of Robotics Level Three. You have learned about robot learning.
You learned that robot learning is the process of a robot improving its behaviour through experience. You learned the difference between programming and learning. You learned about reinforcement learning, Markov Decision Processes, rewards, and penalties.
You learned about Q-Learning and SARSA. You learned about deep reinforcement learning. You learned about imitation learning, sim-to-real transfer, transfer learning, and active learning.
You learned about real robots and their learning methods. You learned about debugging robot learning systems. You learned about the future of robot learning.
You also learned many examples from Nigeria, from your home, from school, and from everyday life. You learned through stories, illustrations, and activities.
Now you are ready for Module Five: Manipulation and Grasping. In Module Five, you will learn how robots pick up and move objects. You will learn about contact modeling, grasping, dexterous manipulation, and force control.
But before you move on, take a moment to review this module. Make sure you understand the key ideas. Practise designing reward functions. Draw diagrams. Test your understanding. The more you practise, the better you will become.
You are doing great. Keep learning. Keep exploring. Keep building. The world of robotics is waiting for you!
End of Module Four
Next: Module Five β Manipulation and Grasping
Welcome to Module Five of Level Three! In Module One, you learned about kinematics and dynamics. In Module Two, you learned about probabilistic robotics. In Module Three, you learned about motion planning. In Module Four, you learned about robot learning. Now you will learn about manipulation and grasping.
Manipulation is the art of using a robot's hands or grippers to move objects. Grasping is the specific act of holding an object. These are some of the most important skills in robotics. Without manipulation, robots cannot do useful work.
Think about your own hands. You use them every day. You pick up a cup. You open a door. You write with a pen. You tie your shoes. Your hands are incredibly useful. Robots with hands can do many of the same things.
In this module, you will learn how robots pick up objects. You will learn about contact modeling, grasping, dexterous manipulation, force control, and multi-fingered hands. These are the tools that make robot arms useful.
Do not worry if these words sound hard. We will explain everything step by step, using simple examples and stories.
Let us begin!
By the end of this module, you will be able to:
Once upon a time, in the city of Kano, Nigeria, there lived a girl named Amina. Amina had built a robot arm named Zaki. Zaki had three joints, a gripper, motors, and a small computer brain.
Amina wanted Zaki to pick up an egg. She thought it would be easy. She programmed Zaki to move to the egg, close the gripper, and lift.
Zaki moved to the egg. The gripper closed. But the gripper closed too hard. The egg cracked. Yolk spilled everywhere.
Amina was frustrated. "Zaki, you broke the egg!" she said.
Zaki did not answer. He just stood there with egg on his gripper.
Amina tried again. This time, she programmed Zaki to close the gripper gently. But the gripper closed too gently. The egg slipped out and fell on the floor. It cracked again.
Amina was even more frustrated. "Zaki, you dropped the egg!" she said.
Amina went to her teacher. "Madam, Zaki cannot pick up an egg. He either breaks it or drops it. What can I do?"
Her teacher smiled. "Amina, you need to learn about force control. Zaki needs to feel how much force to use. He needs to be gentle but firm."
"How can Zaki feel?" Amina asked.
"Use a force sensor," her teacher explained. "The force sensor tells Zaki how much force he is applying. He can adjust to hold the egg without breaking it."
Amina was fascinated. She learned about force sensors. She learned about impedance control β a way to control how stiff or soft the robot's grip is. She learned about contact modeling β understanding how objects touch each other.
She added a force sensor to Zaki's gripper. She wrote a new program. The program used impedance control. It adjusted the grip force based on the sensor reading.
She tested the new program. Zaki moved to the egg. The gripper closed gently. The force sensor detected the egg. Zaki adjusted the force. He lifted the egg. He moved it to a plate. He placed it gently. The egg was not broken.
Amina cheered. "Zaki, you did it!" she shouted.
Zaki had learned to manipulate objects with care. He had learned force control.
That is what you will learn in this module. You will learn how robots pick up and move objects. You will learn about contact modeling, grasping, force control, and dexterous manipulation.
Let us begin!
Manipulation is the act of using a robot's hands or grippers to move or change objects. It includes picking up, placing, pushing, pulling, and turning.
Manipulation is what makes robots useful in the real world. Without manipulation, robots can only observe. With manipulation, robots can work. They can assemble products, pack boxes, cook food, and help people.
Think about your hands. You use them to pick up a cup. You use them to open a door. You use them to write. That is manipulation. Robots with hands can do the same things.
A factory robot arm picks up a car part and places it on an assembly line.
A school robot arm picks up a block and stacks it on another block.
A robot vacuum picks up dust and dirt from the floor.
A robot in a Lagos factory picks up bottles and places them in a crate.
Manipulation Object | V +-------+ | Robot | | Hand | +-------+ | V Object moved
Manipulation is the act of using a robot's hands or grippers to move or change objects.
Grasping is the act of holding an object with a robot's hand or gripper. It is a specific type of manipulation.
Grasping is the first step in most manipulation tasks. If the robot cannot grasp an object, it cannot move it. Good grasping is essential for useful robots.
Think about picking up a cup. You wrap your fingers around it. You hold it firmly. That is grasping. A robot uses a gripper to do the same thing.
| Type | Description | Example |
|---|---|---|
| Power Grasp | Whole hand holds object | Holding a hammer |
| Precision Grasp | Fingertips hold object | Holding a pen |
| Pinch Grasp | Two fingers hold object | Picking up a coin |
| Enveloping Grasp | Hand wraps around object | Holding a ball |
A robot arm grasps a bottle and places it in a box.
A robot grasps a pencil and writes on paper.
A robot grasps a plate and places it on a table.
A robot grasps a tomato and places it in a basket.
Grasping Object | V +-------+ |Gripper| | || | | || | +-------+ | V Object held
Grasping is the act of holding an object with a robot's hand or gripper.
Contact modeling is the study of how objects touch each other. It describes the forces and movements when two surfaces meet.
Contact modeling is essential for grasping. Without it, the robot cannot know how much force to use. It might crush the object or drop it.
Think about pressing your finger on a table. You feel resistance. The table pushes back. That is contact. Contact modeling describes this push and pull.
| Type | Description | Example |
|---|---|---|
| Point Contact | Touch at a single point | Ball on a table |
| Line Contact | Touch along a line | Cylinder on a table |
| Surface Contact | Touch over an area | Box on a table |
When you press a button, your finger makes contact. Contact modeling describes the force.
When you write with a pencil, the pencil makes contact with paper.
When you place a cup on a table, the cup makes contact with the table.
When a trader places tomatoes in a basket, the tomatoes make contact with each other.
Contact Modeling
Object 1
+-------+
| |
+-------+
|
| Contact
V
+-------+
| |
+-------+
Object 2
Forces at contact point.
Contact modeling is the study of how objects touch each other. It describes the forces and movements when two surfaces meet.
Prehensile manipulation uses grasping to move objects. Non-prehensile manipulation moves objects without grasping.
Different tasks need different types of manipulation. Some objects are too big to grasp. Some are too small. Knowing the difference helps the robot choose the right approach.
Prehensile: You pick up a cup with your hand. Non-prehensile: You push a box across the floor with your foot.
| Aspect | Prehensile | Non-Prehensile |
|---|---|---|
| Uses grasping? | Yes | No |
| Example | Picking up a cup | Pushing a box |
| Best for | Small objects | Large objects |
A robot picks up a ball (prehensile). A robot pushes a box (non-prehensile).
You pick up a pencil (prehensile). You push a book across a desk (non-prehensile).
You pick up a plate (prehensile). You push a chair (non-prehensile).
You pick up a tomato (prehensile). You push a wheelbarrow (non-prehensile).
Prehensile vs Non-Prehensile Prehensile: +-------+ |Gripper| | || | +-------+ Object held Non-Prehensile: +-------+ | Robot | +-------+ | V Object pushed
Prehensile manipulation uses grasping to move objects. Non-prehensile manipulation moves objects without grasping.
Grasp quality is a measure of how good a grasp is. A good grasp is stable and secure. A bad grasp is unstable and might drop the object.
Grasp quality determines if the robot can hold the object. A bad grasp leads to dropped objects. A good grasp leads to successful manipulation.
Think about holding a cup. If you hold it by the handle, it is stable. If you hold it by the rim, it might slip. That is grasp quality.
A robot grasps a bottle by the body, not the cap. This gives better grasp quality.
You hold a book with both hands for better stability.
You hold a pot with both hands for better control.
You hold a basket of tomatoes with both hands for balance.
Grasp Quality Good grasp: +-------+ |Gripper| | || | | || | +-------+ Object stable Bad grasp: +-------+ |Gripper| | | | +-------+ Object slipping
Grasp quality is a measure of how good a grasp is. A good grasp is stable and secure.
Dexterous manipulation is the ability to move objects within the hand without releasing them. It requires fine control and many degrees of freedom.
Dexterous manipulation allows robots to do complex tasks. They can rotate objects, adjust their grip, and handle delicate items.
Think about spinning a pen in your fingers. You are moving it without dropping it. That is dexterous manipulation.
A robot hand rotates a screw to tighten it.
You rotate a pencil to use the eraser.
You rotate a key to unlock a door.
A trader rotates a fruit to check all sides for ripeness.
Dexterous Manipulation +-------+ | Hand | | || | | || | +-------+ | V Object rotated within hand
Dexterous manipulation is the ability to move objects within the hand without releasing them.
Operational space control is a way to control a robot arm by specifying the position and force of the hand, not the joint angles.
Operational space control is more natural. We think about where the hand should go, not what the joints should do. It simplifies control.
Think about reaching for a cup. You think about where your hand should go. You do not think about your shoulder and elbow angles. That is operational space control.
A robot arm moves its hand to a target position using operational space control.
You move your hand to write on the board.
You move your hand to pick up a spoon.
You move your hand to pick up a plate of jollof rice.
Operational Space Control
Target position
|
V
+-------------+
| Controller |
+-------------+
|
V
Joint angles
|
V
+-------------+
| Robot Arm |
+-------------+
|
V
Hand at target
Operational space control is a way to control a robot arm by specifying the position and force of the hand.
Force control is a way to control how much force a robot applies. It is used when the robot interacts with objects.
Force control prevents the robot from crushing objects. It allows the robot to handle delicate items.
Think about shaking hands. You do not squeeze too hard. You use just the right amount of force. That is force control.
A robot uses force control to pick up an egg without breaking it.
You use force control to write with a pencil without breaking the tip.
You use force control to hold a glass without breaking it.
You use force control to hold a tomato without crushing it.
Force Control
Force sensor
|
V
+-------------+
| Controller |
+-------------+
|
V
Grip force
|
V
Object held without damage
Force control is a way to control how much force a robot applies. It prevents the robot from crushing objects.
Impedance control is a way to control how stiff or soft a robot's motion is. It makes the robot behave like a spring.
Impedance control is safer. If the robot hits something, it gives way instead of pushing hard. It is used in human-robot interaction.
Think about pushing a door. A stiff door does not move. A soft door moves easily. Impedance control makes the robot stiff or soft.
A robot uses impedance control to shake hands with a human safely.
You use impedance control when playing with a soft ball.
You use impedance control when handling a baby.
You use impedance control when carrying a basket of eggs.
Impedance Control Stiff: Soft: +-------+ +-------+ | Robot | | Robot | +-------+ +-------+ || ~~~ || ~~~ Hard contact Soft contact
Impedance control is a way to control how stiff or soft a robot's motion is.
Grippers are the parts of a robot that grasp objects. There are many types of grippers.
Different objects need different grippers. A gripper for a ball is different from a gripper for a pencil. Choosing the right gripper is important.
| Type | Description | Best For |
|---|---|---|
| Two-Finger Gripper | Two fingers open and close | Simple objects |
| Three-Finger Gripper | Three fingers for better grip | Round objects |
| Multi-Finger Hand | Many fingers like a human hand | Complex tasks |
| Vacuum Gripper | Uses suction to hold objects | Flat objects |
| Magnetic Gripper | Uses magnets to hold objects | Metal objects |
A factory robot uses a vacuum gripper to pick up flat panels.
A school robot uses a two-finger gripper to pick up blocks.
A robot vacuum uses suction to pick up dust.
A robot in a Lagos factory uses a magnetic gripper to pick up metal parts.
Types of Grippers Two-Finger: +-------+ | | | | +-------+ Three-Finger: +-------+ | | | | | | | +-------+ Vacuum: +-------+ | O | +-------+
Grippers are the parts of a robot that grasp objects. Different types are used for different objects.
Grasping unknown objects means picking up objects the robot has never seen before. The robot must figure out how to grasp them.
In the real world, robots encounter new objects all the time. They cannot be programmed for every object. They must learn to grasp unknown objects.
Imagine you are blindfolded. Someone gives you an object. You feel it. You figure out how to hold it. That is grasping unknown objects.
A robot in a warehouse picks up packages of different shapes and sizes.
A school robot picks up objects it has never seen before.
A robot picks up toys of different shapes.
A robot in a market picks up fruits of different shapes and sizes.
Grasping Unknown Objects
Object unknown
|
V
+-------------+
| Sensors |
| (see object)|
+-------------+
|
V
+-------------+
| Decide grasp|
+-------------+
|
V
Grasp and lift
Grasping unknown objects means picking up objects the robot has never seen before.
Multi-fingered hands are robot hands with many fingers. They can do complex grasps and dexterous manipulation.
Multi-fingered hands are more flexible. They can handle many different objects. They are used in advanced robotics.
Think about your hand. It has five fingers. You can hold a cup, a pen, a ball, and many other objects. A multi-fingered robot hand works the same way.
A humanoid robot uses a multi-fingered hand to pick up a cup.
A school robot uses a three-fingered hand to pick up a ball.
A robot toy uses a multi-fingered hand to hold objects.
A robot in a hospital uses a multi-fingered hand to handle surgical tools.
Multi-Fingered Hand +-----------+ | Hand | | || || || | | || || || | | || || || | +-----------+ | | | V V V Multiple fingers
Multi-fingered hands are robot hands with many fingers. They can do complex grasps and dexterous manipulation.
Real robots use manipulation to do useful work.
| Robot | Manipulation Type | Purpose |
|---|---|---|
| Factory Robot Arm | Prehensile, force control | Assemble products |
| Warehouse Robot | Prehensile, grasping unknown objects | Pack boxes |
| Surgical Robot | Dexterous, force control | Perform surgery |
| Humanoid Robot | Multi-fingered, dexterous | Help humans |
| Robot Vacuum | Non-prehensile | Clean floors |
A factory robot arm assembles a phone using force control.
A school robot arm picks up and stacks blocks.
A robot vacuum uses non-prehensile manipulation to clean.
A robot in a Lagos factory picks up bottles and places them in crates.
Real Robot: Factory Robot Arm
+-------------------+
| Robot Arm |
| +-----------+ |
| | Joint 1 | |
| +-----------+ |
| +-----------+ |
| | Joint 2 | |
| +-----------+ |
| +-----------+ |
| | Gripper | |
| +-----------+ |
+-------------------+
|
V
Picks and places objects
Real robots use manipulation to do useful work. Different types of manipulation are used for different tasks.
Debugging manipulation problems means finding and fixing issues in grasping and manipulation.
Manipulation can fail. The robot might drop the object. It might crush it. It might miss the object. Debugging helps you fix these problems.
| Problem | Cause | Solution |
|---|---|---|
| Object slips | Not enough force | Increase grip force |
| Object crushed | Too much force | Use force control |
| Gripper misses object | Wrong position | Adjust position |
| Object rotates unexpectedly | Poor grasp | Improve grasp quality |
| Robot cannot reach object | Kinematic limits | Move robot closer |
If a robot drops a bottle, check the grip force.
If a robot crushes a block, use force control.
If a robot vacuum misses dirt, adjust the suction.
If a robot in a Lagos factory drops a bottle, check the gripper.
Debugging Manipulation
[ Object dropped ]
|
V
[ Check grip force ]
|
V
[ Check position ]
|
V
[ Check gripper ]
|
V
[ Fix problem ]
|
V
[ Test again ]
Debugging manipulation problems means finding and fixing issues in grasping and manipulation.
The future of manipulation is very exciting. Robots will become more dexterous, more capable, and more useful.
Better manipulation means better robots. They will work in more complex environments. They will help people in more ways.
Researchers are already teaching robots to fold clothes.
Students today are learning manipulation for future jobs.
Robot vacuums are becoming smarter every year.
Nigerian universities are researching manipulation for agriculture and healthcare.
Future of Manipulation
Today: Simple grasping
|
V
Soon: Dexterous manipulation
|
V
Future: Human-like hands
|
V
Future: Robots that cook and clean
The future of manipulation is exciting. Robots will become more dexterous and more capable.
| Word | Simple Definition |
|---|---|
| Manipulation | Using a robot's hands to move or change objects. |
| Grasping | Holding an object with a robot's hand or gripper. |
| Contact Modeling | Studying how objects touch each other. |
| Prehensile | Using grasping to move objects. |
| Non-Prehensile | Moving objects without grasping. |
| Grasp Quality | A measure of how good a grasp is. |
| Dexterous Manipulation | Moving objects within the hand without releasing them. |
| Operational Space Control | Controlling the hand position and force directly. |
| Force Control | Controlling how much force a robot applies. |
| Impedance Control | Controlling how stiff or soft a robot's motion is. |
| Gripper | The part of a robot that grasps objects. |
| Multi-Fingered Hand | A robot hand with many fingers. |
| Concept | Real-Life Example |
|---|---|
| Manipulation | Factory robot picks up car parts. |
| Grasping | Robot arm grasps a bottle. |
| Contact Modeling | Pressing a button. |
| Prehensile | Picking up a cup. |
| Non-Prehensile | Pushing a box. |
| Grasp Quality | Holding a bottle by the body. |
| Dexterous Manipulation | Spinning a pen in your fingers. |
| Operational Space Control | Moving your hand to a target. |
| Force Control | Picking up an egg without breaking it. |
| Impedance Control | Shaking hands gently. |
| Grippers | Vacuum gripper picks up flat panels. |
| Multi-Fingered Hands | Humanoid robot picks up a cup. |
| Concept | Nigerian Example |
|---|---|
| Manipulation | Robot in a Lagos factory picks up bottles. |
| Grasping | Robot grasps a tomato. |
| Contact Modeling | Placing tomatoes in a basket. |
| Prehensile | Picking up a tomato. |
| Non-Prehensile | Pushing a wheelbarrow. |
| Grasp Quality | Holding a basket of tomatoes with both hands. |
| Dexterous Manipulation | Rotating a fruit to check ripeness. |
| Operational Space Control | Moving your hand to pick up jollof rice. |
| Force Control | Holding a tomato without crushing it. |
| Impedance Control | Carrying a basket of eggs. |
| Grippers | Magnetic gripper picks up metal parts. |
| Multi-Fingered Hands | Hospital robot handles surgical tools. |
| Concept | Everyday Example |
|---|---|
| Manipulation | Picking up a cup. |
| Grasping | Holding a spoon. |
| Contact Modeling | Pressing a light switch. |
| Prehensile | Picking up a plate. |
| Non-Prehensile | Pushing a chair. |
| Grasp Quality | Holding a pot with both hands. |
| Dexterous Manipulation | Rotating a key to unlock a door. |
| Operational Space Control | Moving your hand to pick up a spoon. |
| Force Control | Holding a glass without breaking it. |
| Impedance Control | Handling a baby gently. |
| Grippers | A robot vacuum uses suction. |
| Multi-Fingered Hands | Your hand holds many objects. |
| Mistake | Why It Is Wrong | How to Fix It |
|---|---|---|
| Too much force | Object is crushed. | Use force control. |
| Too little force | Object slips. | Increase grip force. |
| Wrong grasp points | Object is unstable. | Improve grasp planning. |
| Ignoring object shape | Grasp fails. | Use vision or touch. |
| Not checking grasp quality | Object drops. | Check grasp before lifting. |
| Not debugging | Problems continue. | Find and fix problems. |
+-------------------+
| Sensors |
| (see object) |
+-------------------+
|
V
+-------------------+
| Grasp Planner |
| (decide grasp) |
+-------------------+
|
V
+-------------------+
| Controller |
| (move gripper) |
+-------------------+
|
V
+-------------------+
| Gripper |
| (grasp object) |
+-------------------+
|
V
+-------------------+
| Object moved |
+-------------------+
( Start )
|
V
+----------------+
| Detect object |
+----------------+
|
V
+----------------+
| Estimate shape |
+----------------+
|
V
+----------------+
| Choose grasp |
+----------------+
|
V
+----------------+
| Move gripper |
+----------------+
|
V
+----------------+
| Close gripper |
+----------------+
|
V
+----------------+
| Check grasp |
+----------------+
/ \
GOOD BAD
/ \
V V
+---------+ +-----------+
| Lift | | Adjust |
| object | | grasp |
+---------+ +-----------+
| Gripper | Best For | Advantage | Disadvantage |
|---|---|---|---|
| Two-Finger | Simple objects | Simple, cheap | Limited grasp types |
| Three-Finger | Round objects | Better grip | More complex |
| Multi-Finger | Complex tasks | Very flexible | Expensive, complex |
| Vacuum | Flat objects | No marks | Only flat surfaces |
| Magnetic | Metal objects | Strong grip | Only magnetic objects |
Step 1: Detect object
|
V
Step 2: Plan grasp
|
V
Step 3: Move gripper
|
V
Step 4: Grasp object
|
V
Step 5: Check grasp
|
V
Step 6: Lift object
|
V
Step 7: Move to destination
|
V
Step 8: Place object
Manipulation is the act of using a robot's hands or grippers to move or change objects.
Grasping is the act of holding an object with a robot's hand or gripper.
Contact modeling is the study of how objects touch each other.
Prehensile manipulation uses grasping. Non-prehensile manipulation moves objects without grasping.
Grasp quality is a measure of how good a grasp is. A good grasp is stable and secure.
Dexterous manipulation is the ability to move objects within the hand without releasing them.
Operational space control is a way to control a robot arm by specifying the position and force of the hand.
Force control is a way to control how much force a robot applies.
Impedance control is a way to control how stiff or soft a robot's motion is.
Grippers are the parts of a robot that grasp objects. Different types are used for different objects.
Grasping unknown objects means picking up objects the robot has never seen before.
Multi-fingered hands are robot hands with many fingers. They can do complex grasps and dexterous manipulation.
Real robots use manipulation to do useful work.
Debugging manipulation problems means finding and fixing issues in grasping and manipulation.
The future of manipulation is exciting. Robots will become more dexterous and more capable.
In this module, you learned about manipulation and grasping. You learned that manipulation is the act of using a robot's hands or grippers to move or change objects. You learned that grasping is the act of holding an object.
You learned about contact modeling β studying how objects touch. You learned about prehensile and non-prehensile manipulation. You learned about grasp quality. You learned about dexterous manipulation.
You learned about operational space control. You learned about force control and impedance control. You learned about different types of grippers. You learned about grasping unknown objects. You learned about multi-fingered hands.
You learned about real robots and their manipulation. You learned about debugging manipulation problems. You learned about the future of manipulation.
Most importantly, you learned that manipulation is what allows robots to do useful work in the real world.
In the next module, you will learn about Human-Robot Interaction. You will learn how humans and robots work together. You will learn about shared control, safety, and cognitive robotics.
But for now, take a moment to celebrate what you have learned. You have taken another big step in your journey to becoming a robotics expert. Well done!
Match the term on the left with its definition on the right.
| Term | Definition |
|---|---|
| 1. Manipulation | A. Holding an object with a robot's hand |
| 2. Grasping | B. Using a robot's hands to move objects |
| 3. Contact Modeling | C. Moving objects within the hand |
| 4. Grasp Quality | D. Studying how objects touch |
| 5. Dexterous Manipulation | E. Controlling how much force is applied |
| 6. Force Control | F. A measure of how good a grasp is |
| 7. Impedance Control | G. The part that grasps objects |
| 8. Gripper | H. Controlling how stiff or soft motion is |
Answers: 1-B, 2-A, 3-D, 4-F, 5-C, 6-E, 7-H, 8-G
Title: Design a Gripper
Instructions:
Example:
Object: Egg Gripper design: - Two soft fingers - Force sensor in each finger - Curved shape to match egg How it works: - Fingers close gently - Force sensor detects egg - Grip force adjusted to hold without breaking
Title: Grasping Scavenger Hunt
Instructions:
Example:
| Object | Grasp Type | Why |
|---|---|---|
| Cup | Power grasp | Handles are easy to hold |
| Pencil | Precision grasp | Need fine control |
| Ball | Enveloping grasp | Round shape |
| Coin | Pinch grasp | Small and flat |
| Bottle | Power grasp | Cylindrical shape |
Title: Build a Simple Gripper
Goal: Create a gripper that can pick up a small object.
Steps:
Deliverables:
Title: Implement Force Control in Simulation
Instructions:
Grading Criteria:
| Criteria | Points |
|---|---|
| Force control works correctly | 30 |
| Egg is picked up without breaking | 25 |
| Multiple force limits tested | 15 |
| Report is clear | 15 |
| Visualization is clear | 15 |
| Total | 100 |
In Module Six, you will learn about Human-Robot Interaction. You will learn how humans and robots work together.
You will learn about:
To prepare for Module Six:
Get ready for an exciting journey into the world of human-robot interaction!
Congratulations! You have completed Module Five of Fundamentals of Robotics Level Three. You have learned about manipulation and grasping.
You learned that manipulation is the act of using a robot's hands or grippers to move or change objects. You learned that grasping is the act of holding an object. You learned about contact modeling, prehensile and non-prehensile manipulation, and grasp quality.
You learned about dexterous manipulation. You learned about operational space control, force control, and impedance control. You learned about different types of grippers. You learned about grasping unknown objects and multi-fingered hands.
You learned about real robots and their manipulation. You learned about debugging manipulation problems. You learned about the future of manipulation.
You also learned many examples from Nigeria, from your home, from school, and from everyday life. You learned through stories, illustrations, and activities.
Now you are ready for Module Six: Human-Robot Interaction. In Module Six, you will learn how humans and robots work together. You will learn about shared control, teleoperation, haptics, safety, and cognitive robotics.
But before you move on, take a moment to review this module. Make sure you understand the key ideas. Practise grasping and manipulation. Draw diagrams. Test your understanding. The more you practise, the better you will become.
You are doing great. Keep learning. Keep exploring. Keep building. The world of robotics is waiting for you!
End of Module Five
Next: Module Six β Human-Robot Interaction
Welcome to Module Seven of Level Three! You have learned so much. You learned about kinematics, dynamics, probabilistic robotics, motion planning, robot learning, manipulation, and human-robot interaction. Now you will learn about advanced application domains.
Advanced application domains are special areas of robotics. Each area has its own challenges and solutions. In this module, you will learn about aerial robotics (drones), locomotion (walking robots), swarm robotics (many robots working together), soft robotics (flexible robots), humanoids and exoskeletons, and more.
Think about all the different robots in the world. Some fly. Some walk. Some swim. Some work in groups. Some are soft and flexible. Some look like humans. Each type of robot is designed for a specific purpose.
In this module, you will explore these exciting areas. You will learn how drones fly, how robots walk, how swarms work, and how soft robots move. You will learn about robots that help in surgery, space, and underwater exploration.
Do not worry if these words sound hard. We will explain everything step by step, using simple examples and stories.
Let us begin!
By the end of this module, you will be able to:
Once upon a time, in the city of Abuja, Nigeria, there was a big Robotics Exhibition. Students from all over the country came to show their robots. There were robots that flew. There were robots that walked. There were robots that swam. There were robots that worked in groups. There were even robots that were soft and flexible.
A group of students from Enugu, led by a girl named Ada, came to the exhibition. They had built a drone that could deliver medicine to remote villages.
Another group from Kano, led by a boy named Tunde, had built a walking robot that could climb stairs.
A third group from Lagos, led by a girl named Ngozi, had built a swarm of small robots that could clean a beach.
A fourth group from Ibadan, led by a boy named Chidi, had built a soft robot that could squeeze through small spaces.
All the students were excited. They walked around the exhibition, looking at each other's robots. They asked questions. They learned from each other.
Ada looked at Tunde's walking robot. "How does it walk without falling?" she asked.
Tunde explained. "It uses locomotion. It has sensors that help it balance. It has motors that move its legs. It uses a special algorithm called a Central Pattern Generator."
Ngozi looked at Chidi's soft robot. "How does it squeeze through small spaces?" she asked.
Chidi explained. "It is made of soft materials. It does not have hard joints. It can bend and twist like a rubber hose."
Ada looked at Ngozi's swarm robots. "How do they work together without bumping into each other?" she asked.
Ngozi explained. "They use swarm intelligence. They follow simple rules. They communicate with each other. Together, they do complex tasks."
The students spent the whole day learning from each other. They realized that robotics is a big field with many exciting areas.
That is what you will learn in this module. You will learn about aerial robotics, locomotion, swarm robotics, soft robotics, humanoids, exoskeletons, and more.
Let us begin!
Aerial robotics is the study of robots that fly. These robots are called drones or unmanned aerial vehicles (UAVs).
Aerial robots can go where humans cannot. They can fly over mountains, into disaster zones, and across oceans. They are used for delivery, photography, search and rescue, and military operations.
Think about a bird flying in the sky. A drone flies the same way. It has spinning blades called propellers. The propellers push air down, and the drone goes up.
| Type | Description | Example |
|---|---|---|
| Multirotor | Has multiple rotors (propellers) | Quadcopter (4 rotors) |
| Fixed-Wing | Has wings like an aeroplane | Long-distance drones |
| Helicopter | Has one main rotor | Heavy-lift drones |
| Hybrid | Combines rotors and wings | VTOL drones |
Amazon uses drones to deliver packages to customers.
A school project might use a small drone to take aerial photos.
A toy drone flies around your backyard.
A Nigerian startup uses drones to deliver medicine to remote villages.
Aerial Robot (Quadcopter)
Rotor 1
|
V
+---------+
| |
| Drone |
| |
+---------+
^
|
Rotor 2
Rotors spin to create lift.
Aerial robotics is the study of robots that fly. They are called drones or UAVs.
Drone flight is based on the principle of lift. The propellers push air down, and the drone goes up.
Understanding flight helps us control drones. We can make them go up, down, forward, backward, and sideways.
Think about a balloon. When you let it go, it flies up. The air inside pushes down. The balloon goes up. A drone works the same way, but with spinning propellers.
| Movement | How It Works |
|---|---|
| Up (Throttle) | Increase rotor speed |
| Down (Throttle) | Decrease rotor speed |
| Forward (Pitch) | Tilt front down |
| Backward (Pitch) | Tilt front up |
| Left (Roll) | Tilt left down |
| Right (Roll) | Tilt right down |
| Turn (Yaw) | Spin rotors in different directions |
A drone pilot uses a controller to move the drone up, down, and around.
Students learn about lift and thrust in physics class.
A toy drone flies around your room.
A drone delivers medicine by flying over roads and rivers.
Drone Flight Lift ^ | +--- Rotor | V Thrust Lift must be greater than weight to go up.
Drones fly using lift. The propellers push air down, and the drone goes up.
Locomotion is the ability to move from one place to another. In robotics, it means how a robot moves around.
Locomotion is essential for robots. Without it, robots cannot explore, work, or help people. Different robots use different types of locomotion.
Think about how you move. You walk. You run. You jump. You swim. That is locomotion. Robots also move in different ways.
| Type | How It Works | Example |
|---|---|---|
| Wheeled | Uses wheels to move | Robot vacuum |
| Legged | Uses legs to walk | Boston Dynamics Spot |
| Tracked | Uses tracks like a tank | Mars rover |
| Flying | Uses wings or rotors | Drone |
| Swimming | Uses fins or propellers | Underwater robot |
A car uses wheeled locomotion. A bird uses flying locomotion.
A robot in a science fair might use wheeled locomotion.
A robot vacuum uses wheeled locomotion to clean.
A keke napep uses wheeled locomotion.
Types of Locomotion Wheeled: +-------+ | Robot | +-------+ O O Legged: +-------+ | Robot | +-------+ | | Flying: +-------+ | Robot | +-------+ \ / V
Locomotion is the ability to move from one place to another. Different robots use different types of locomotion.
Walking robots are robots that use legs to move. They can walk on uneven ground and climb stairs.
Walking robots can go where wheeled robots cannot. They can climb stairs, walk on rocks, and navigate rough terrain.
Think about how you walk. You lift one leg, move it forward, and put it down. Then you do the same with the other leg. Walking robots do the same thing.
Boston Dynamics has a robot called Spot that walks on four legs.
A school robot might use two legs to walk.
A robot toy might walk on two legs.
A robot that climbs stairs might be used in a Nigerian hospital.
Walking Robot
+-------+
| Robot |
+-------+
| |
| |
/ \
/ \
Legs move to walk.
Walking robots use legs to move. They can climb stairs and navigate rough terrain.
Swarm robotics is the study of many robots working together. They follow simple rules and communicate with each other.
Swarm robots can do tasks that are too big for one robot. They can cover large areas, search for survivors, and clean beaches.
Think about ants. One ant is small. But many ants can build a nest, find food, and defend their colony. Swarm robots work the same way.
A swarm of drones can search a large area for a missing person.
A group of robots can clean a classroom together.
A swarm of small robots can clean a house.
A swarm of robots can clean a beach in Lagos.
Swarm Robotics
Robot 1 <--> Robot 2
| |
| |
V V
Robot 3 <--> Robot 4
| |
+------------+
|
V
Task completed
Swarm robotics is the study of many robots working together. They follow simple rules and communicate with each other.
Soft robotics is the study of robots made from soft, flexible materials. They can bend, twist, and squeeze.
Soft robots are safer around humans. They can squeeze through small spaces. They can handle delicate objects. They are used in surgery and manufacturing.
Think about a rubber hose. It can bend and twist. A soft robot is like a rubber hose. It can move in ways that hard robots cannot.
| Material | Property | Use |
|---|---|---|
| Silicone | Flexible, soft | Soft grippers |
| Rubber | Stretchy, durable | Soft actuators |
| Hydrogel | Water-based, soft | Medical robots |
| Shape Memory Alloy | Changes shape with heat | Soft muscles |
A soft robot can squeeze through a small gap to reach a trapped person.
A school project might use silicone to make a soft gripper.
A soft robot toy can bend and twist.
A soft robot can pick up tomatoes without crushing them.
Soft Robotics Hard Robot: +-------+ | | +-------+ Soft Robot: +~~~~~~~+ | | +~~~~~~~+ Soft robot can bend and twist.
Soft robotics is the study of robots made from soft, flexible materials. They can bend, twist, and squeeze.
Humanoids are robots that look like humans. Exoskeletons are wearable robots that help humans move.
Humanoids can work in environments designed for humans. Exoskeletons can help people with disabilities walk and lift heavy objects.
Think about a robot that looks like you. It has a head, arms, and legs. That is a humanoid. Think about a suit that helps you lift heavy things. That is an exoskeleton.
| Type | Description | Example |
|---|---|---|
| Humanoid Robot | Looks like a human | ASIMO, Atlas |
| Upper-Body Exoskeleton | Helps arms and shoulders | Factory workers |
| Lower-Body Exoskeleton | Helps legs and hips | Walking assistance |
| Full-Body Exoskeleton | Helps whole body | Heavy lifting |
A humanoid robot can greet customers in a hotel.
A school robot might look like a human and answer questions.
An exoskeleton can help an elderly person walk.
An exoskeleton can help a Nigerian worker lift heavy loads safely.
Humanoid Robot
+---+
| O |
+---+
/ | \
/ | \
/ | \
/ | \
+-----+-----+
| | |
| | |
+-----+-----+
| |
| |
/ \
/ \
Looks like a human.
Humanoids are robots that look like humans. Exoskeletons are wearable robots that help humans move.
Underwater robotics is the study of robots that work underwater. They are used for exploration, research, and repair.
Underwater robots can go deeper than humans. They can explore the ocean, repair pipelines, and study marine life.
Think about a submarine. It can go underwater and explore. An underwater robot works the same way, but without people inside.
| Type | Description | Use |
|---|---|---|
| ROV | Remotely Operated Vehicle | Repair, inspection |
| AUV | Autonomous Underwater Vehicle | Exploration, mapping |
| Glider | Uses buoyancy to move | Long-term monitoring |
Underwater robots explore shipwrecks and coral reefs.
A school project might build a small underwater robot.
A pool robot cleans the bottom of a swimming pool.
An underwater robot inspects oil pipelines in the Niger Delta.
Underwater Robot ~~~~~~~~~~~~~ ~ ~ ~ +-----+ ~ ~ |Robot| ~ ~ +-----+ ~ ~ ~ ~~~~~~~~~~~~~ Works underwater.
Underwater robotics is the study of robots that work underwater. They are used for exploration, research, and repair.
Space robotics is the study of robots that work in space. They explore planets, repair satellites, and build structures.
Space is dangerous for humans. Robots can work in space without risking human life. They can explore distant planets and gather data.
Think about a robot on Mars. It drives around, takes pictures, and studies rocks. That is space robotics.
| Robot | Mission | Purpose |
|---|---|---|
| Mars Rover | Explore Mars | Study rocks and soil |
| Canadarm | International Space Station | Repair and build |
| Voyager | Explore outer space | Study planets |
NASA's Perseverance rover is exploring Mars right now.
A school project might build a model Mars rover.
A toy robot might simulate space exploration.
Nigerian students can learn about space robotics and dream of joining space agencies.
Space Robot
* * *
* *
* +--+ *
* | | *
* +--+ *
* / \ *
* / \ *
* *
* * *
Robot explores space.
Space robotics is the study of robots that work in space. They explore planets, repair satellites, and build structures.
Surgical robotics is the study of robots that help doctors perform surgery. Medical robotics is the study of robots that help in healthcare.
Surgical robots are more precise than human hands. They can make tiny incisions. They help patients recover faster.
Think about a surgeon using a robot to perform surgery. The robot holds the tools. The surgeon controls the robot. That is surgical robotics.
| Robot | Use | Benefit |
|---|---|---|
| Da Vinci | Surgery | Precise movements |
| Rehabilitation Robot | Physical therapy | Helps patients recover |
| Hospital Delivery Robot | Deliver medicine | Saves time |
A surgical robot helps a surgeon perform a delicate operation.
A school project might build a model surgical robot.
A robot might help an elderly person take medicine.
A Nigerian hospital uses a surgical robot to perform precise operations.
Surgical Robot Surgeon | V +-------+ | Robot | +-------+ | V Patient Robot helps surgeon.
Surgical robotics is the study of robots that help doctors perform surgery. Medical robotics is the study of robots that help in healthcare.
Agricultural robotics is the study of robots that help farmers. They plant, water, and harvest crops.
Agricultural robots can work faster and more efficiently than humans. They can help farmers grow more food.
Think about a robot that plants seeds. It moves along a field, digs a hole, drops a seed, and covers it. That is agricultural robotics.
| Robot | Use | Benefit |
|---|---|---|
| Planting Robot | Plants seeds | Faster planting |
| Harvesting Robot | Picks crops | Less waste |
| Weeding Robot | Removes weeds | No chemicals |
| Monitoring Robot | Checks crop health | Early detection |
A robot picks apples from a tree without bruising them.
A school project might build a robot that waters plants.
A robot might water your garden.
A Nigerian farmer uses a robot to plant yams.
Agricultural Robot
Field
+-------------------+
| O O O O O O O O |
| O O O O O O O O |
| O O O O O O O O |
+-------------------+
^
|
Robot
Robot plants or harvests crops.
Agricultural robotics is the study of robots that help farmers. They plant, water, and harvest crops.
Sustainable robotics is the study of robots that are good for the environment. They use less energy and produce less waste.
The world needs to protect the environment. Sustainable robots help by using clean energy and reducing pollution.
Think about a robot that uses solar power. It does not need batteries. It does not pollute. That is sustainable robotics.
| Robot | Feature | Benefit |
|---|---|---|
| Solar-Powered Robot | Uses solar energy | No pollution |
| Recycling Robot | Sorts waste | Less landfill |
| Tree-Planting Robot | Plants trees | Fights climate change |
A robot sorts plastic bottles for recycling.
A school project might build a robot that sorts waste.
A robot might help you recycle at home.
A robot in Lagos sorts plastic waste for recycling.
Sustainable Robotics
Solar Panel
|
V
+-------+
| Robot |
+-------+
|
V
Clean work
Uses clean energy.
Sustainable robotics is the study of robots that are good for the environment.
Real robots work in many advanced domains.
| Robot | Domain | Purpose |
|---|---|---|
| DJI Phantom | Aerial | Photography, delivery |
| Boston Dynamics Spot | Locomotion | Inspection, rescue |
| Kilobots | Swarm | Research |
| Octobot | Soft | Research |
| Atlas | Humanoid | Research |
| Da Vinci | Surgical | Surgery |
| Perseverance | Space | Mars exploration |
Boston Dynamics Spot inspects oil rigs and construction sites.
Students learn about these robots in robotics class.
Robot vacuums are a common example of advanced robotics in the home.
Nigerian students can build robots for agriculture, healthcare, and environmental monitoring.
Real Robots
Aerial: +---+
| D |
+---+
/ \
Legged: +---+
| S |
+---+
| |
Swarm: * * *
* * *
Soft: +~~~+
| |
+~~~+
Real robots work in many advanced domains. Each robot is designed for a specific purpose.
Debugging advanced robotic systems means finding and fixing problems in complex robots.
Advanced robots are complex. Many things can go wrong. Debugging helps you find and fix these problems.
| Problem | Domain | Solution |
|---|---|---|
| Drone crashes | Aerial | Check rotors and battery |
| Walking robot falls | Locomotion | Check balance sensors |
| Swarm robots collide | Swarm | Improve communication |
| Soft robot tears | Soft | Use stronger material |
| Surgical robot inaccurate | Surgical | Calibrate sensors |
If a drone crashes, check the rotors and battery.
If a walking robot falls, check the balance sensors.
If a robot vacuum gets stuck, check the sensors.
If a delivery drone crashes, check the GPS and rotors.
Debugging Advanced Robots
[ Robot fails ]
|
V
[ Check sensors ]
|
V
[ Check motors ]
|
V
[ Check software ]
|
V
[ Fix problem ]
|
V
[ Test again ]
Debugging advanced robotic systems means finding and fixing problems in complex robots.
The future of advanced robotics is very exciting. Robots will become more capable, more intelligent, and more useful.
Better robots will help people in more ways. They will work in more places. They will solve more problems.
Researchers are already building robots that can walk, swim, and fly.
Students today are learning advanced robotics for future jobs.
Advanced robots will become common in homes.
Nigerian universities are researching advanced robotics for agriculture, healthcare, and environmental monitoring.
Future of Advanced Robotics
Today: Simple robots
|
V
Soon: Advanced robots
|
V
Future: Robots in every domain
|
V
Future: Robots help everyone
The future of advanced robotics is exciting. Robots will become more capable and more useful.
| Word | Simple Definition |
|---|---|
| Aerial Robotics | Robots that fly. |
| Drone | An unmanned aerial vehicle. |
| Locomotion | Moving from one place to another. |
| Walking Robot | A robot that uses legs to move. |
| Swarm Robotics | Many robots working together. |
| Soft Robotics | Robots made from soft, flexible materials. |
| Humanoid | A robot that looks like a human. |
| Exoskeleton | A wearable robot that helps humans move. |
| Underwater Robotics | Robots that work underwater. |
| Space Robotics | Robots that work in space. |
| Surgical Robotics | Robots that help doctors perform surgery. |
| Agricultural Robotics | Robots that help farmers. |
| Sustainable Robotics | Robots that are good for the environment. |
| Concept | Real-Life Example |
|---|---|
| Aerial Robotics | Amazon uses drones to deliver packages. |
| Drone Flight | Drone pilot uses controller. |
| Locomotion | Car uses wheeled locomotion. |
| Walking Robots | Boston Dynamics Spot walks on four legs. |
| Swarm Robotics | Swarm of drones searches for missing person. |
| Soft Robotics | Soft robot squeezes through small gap. |
| Humanoids | Humanoid robot greets customers. |
| Exoskeletons | Exoskeleton helps elderly person walk. |
| Underwater Robotics | Underwater robot explores shipwreck. |
| Space Robotics | Perseverance rover explores Mars. |
| Surgical Robotics | Da Vinci robot helps surgeon. |
| Agricultural Robotics | Robot picks apples without bruising. |
| Sustainable Robotics | Solar-powered robot sorts recycling. |
| Concept | Nigerian Example |
|---|---|
| Aerial Robotics | Drone delivers medicine to remote villages. |
| Drone Flight | Drone flies over roads and rivers. |
| Locomotion | Keke napep uses wheeled locomotion. |
| Walking Robots | Robot climbs stairs in a Nigerian hospital. |
| Swarm Robotics | Swarm of robots cleans a beach in Lagos. |
| Soft Robotics | Soft robot picks up tomatoes without crushing. |
| Humanoids | Humanoid robot greets patients in a Lagos hospital. |
| Exoskeletons | Exoskeleton helps worker lift heavy loads. |
| Underwater Robotics | Underwater robot inspects oil pipelines in Niger Delta. |
| Space Robotics | Nigerian students learn about space robotics. |
| Surgical Robotics | Nigerian hospital uses surgical robot. |
| Agricultural Robotics | Nigerian farmer uses robot to plant yams. |
| Sustainable Robotics | Robot in Lagos sorts plastic waste. |
| Concept | Everyday Example |
|---|---|
| Aerial Robotics | Drone photography at weddings. |
| Drone Flight | Toy drone in the park. |
| Locomotion | Bicycle uses wheeled locomotion. |
| Walking Robots | Robot toy walks on two legs. |
| Swarm Robotics | Bees working together. |
| Soft Robotics | Rubber hose bends. |
| Humanoids | Robot at a science museum. |
| Exoskeletons | Knee brace helps you walk. |
| Underwater Robotics | Pool cleaner robot. |
| Space Robotics | Documentary about Mars rover. |
| Surgical Robotics | Hospital uses robot for surgery. |
| Agricultural Robotics | Robot waters garden. |
| Sustainable Robotics | Robot sorts recycling. |
| Mistake | Why It Is Wrong | How to Fix It |
|---|---|---|
| Not checking drone battery | Drone crashes. | Check battery before flight. |
| Not balancing walking robot | Robot falls. | Check balance sensors. |
| Swarm robots collide | Task fails. | Improve communication. |
| Soft robot tears | Robot damaged. | Use stronger material. |
| Surgical robot inaccurate | Patient harmed. | Calibrate sensors. |
| Not debugging | Problems continue. | Find and fix problems. |
+-------------------+
| Advanced |
| Robotics |
+-------------------+
| | | |
| | | |
V V V V
+-----+ +-----+ +-----+ +-----+
|Aerial| |Walk | |Swarm| |Soft |
+-----+ +-----+ +-----+ +-----+
| | | |
V V V V
+-----+ +-----+ +-----+ +-----+
|Under| |Space| |Surg | |Agri |
|water| | | |ical | |cult |
+-----+ +-----+ +-----+ +-----+
( Start )
|
V
+----------------+
| Identify domain|
+----------------+
|
V
+----------------+
| Choose |
| locomotion |
+----------------+
|
V
+----------------+
| Choose |
| materials |
+----------------+
|
V
+----------------+
| Design sensors |
+----------------+
|
V
+----------------+
| Design control |
+----------------+
|
V
+----------------+
| Test and |
| improve |
+----------------+
|
V
( Deploy )
| Domain | Key Feature | Example |
|---|---|---|
| Aerial | Flying | Drone |
| Locomotion | Walking | Spot |
| Swarm | Teamwork | Kilobots |
| Soft | Flexible | Octobot |
| Humanoid | Human-like | Atlas |
| Underwater | Deep water | ROV |
| Space | Other planets | Perseverance |
| Surgical | Precision | Da Vinci |
| Agricultural | Farming | Harvesting robot |
| Sustainable | Eco-friendly | Solar robot |
Step 1: Identify need
|
V
Step 2: Choose domain
|
V
Step 3: Design robot
|
V
Step 4: Build prototype
|
V
Step 5: Test
|
V
Step 6: Debug
|
V
Step 7: Improve
|
V
Step 8: Deploy
Aerial robotics is the study of robots that fly. They are called drones or UAVs.
Drones fly using lift. The propellers push air down, and the drone goes up.
Locomotion is the ability to move from one place to another. Different robots use different types of locomotion.
Walking robots use legs to move. They can climb stairs and navigate rough terrain.
Swarm robotics is the study of many robots working together.
Soft robotics is the study of robots made from soft, flexible materials.
Humanoids are robots that look like humans. Exoskeletons are wearable robots that help humans move.
Underwater robotics is the study of robots that work underwater.
Space robotics is the study of robots that work in space.
Surgical robotics is the study of robots that help doctors perform surgery.
Agricultural robotics is the study of robots that help farmers.
Sustainable robotics is the study of robots that are good for the environment.
Real robots work in many advanced domains. Each robot is designed for a specific purpose.
Debugging advanced robotic systems means finding and fixing problems in complex robots.
The future of advanced robotics is exciting. Robots will become more capable and more useful.
In this module, you learned about advanced application domains. You learned about aerial robotics β robots that fly. You learned about locomotion β how robots move. You learned about walking robots β robots that use legs.
You learned about swarm robotics β many robots working together. You learned about soft robotics β robots made from flexible materials. You learned about humanoids and exoskeletons. You learned about underwater robotics and space robotics. You learned about surgical robotics, agricultural robotics, and sustainable robotics.
You learned about real robots in advanced domains. You learned about debugging advanced robotic systems. You learned about the future of advanced robotics.
Most importantly, you learned that robotics is a big field with many exciting areas. Each area has its own challenges and solutions.
In the next module, you will complete a Capstone Project. You will put everything you have learned into one big project.
But for now, take a moment to celebrate what you have learned. You have taken another big step in your journey to becoming a robotics expert. Well done!
Match the domain on the left with its description on the right.
| Domain | Description |
|---|---|
| 1. Aerial Robotics | A. Many robots working together |
| 2. Locomotion | B. Robots that fly |
| 3. Swarm Robotics | C. Robots that work underwater |
| 4. Soft Robotics | D. Moving from one place to another |
| 5. Underwater Robotics | E. Robots made from flexible materials |
| 6. Space Robotics | F. Robots that help farmers |
| 7. Surgical Robotics | G. Robots that work in space |
| 8. Agricultural Robotics | H. Robots that help doctors |
Answers: 1-B, 2-D, 3-A, 4-E, 5-C, 6-G, 7-H, 8-F
Title: Design an Advanced Robot
Instructions:
Example:
Domain: Agricultural Robotics Robot: Yam planting robot Design: - Wheels for locomotion - Digging tool - Seed holder - Sensor to detect soil How it works: 1. Robot moves along field 2. Sensor detects soil 3. Digging tool makes hole 4. Seed dropped in hole 5. Soil covered 6. Robot moves to next spot
Title: Advanced Robotics Scavenger Hunt
Instructions:
Example:
| Device | Domain | Purpose |
|---|---|---|
| Drone | Aerial | Photography |
| Robot vacuum | Locomotion | Cleaning |
| Pool cleaner | Underwater | Cleaning pool |
| Solar robot | Sustainable | Recycling |
| Robot toy | Humanoid | Play |
Title: Build a Simple Advanced Robot
Goal: Create a robot that demonstrates an advanced application.
Steps:
Deliverables:
Title: Implement a Swarm Robotics Simulation
Instructions:
Grading Criteria:
| Criteria | Points |
|---|---|
| Swarm works together | 30 |
| Task is completed | 25 |
| Communication works | 15 |
| Report is clear | 15 |
| Visualization is clear | 15 |
| Total | 100 |
In Module Eight, you will complete a Capstone Project. You will put everything you have learned into one big project.
To prepare for Module Eight:
Get ready for an exciting journey into the world of complete robot projects!
Congratulations! You have completed Module Seven of Fundamentals of Robotics Level Three. You have learned about advanced application domains.
You learned about aerial robotics, locomotion, walking robots, swarm robotics, soft robotics, humanoids, exoskeletons, underwater robotics, space robotics, surgical robotics, agricultural robotics, and sustainable robotics.
You learned about real robots in advanced domains. You learned about debugging advanced robotic systems. You learned about the future of advanced robotics.
You also learned many examples from Nigeria, from your home, from school, and from everyday life. You learned through stories, illustrations, and activities.
Now you are ready for Module Eight: Capstone Project. In Module Eight, you will put everything you have learned into one big project. You will design, build, program, and test a complete robot. You will present your project to the class.
But before you move on, take a moment to review this module. Make sure you understand the key ideas. Practise designing advanced robots. Draw diagrams. Test your understanding. The more you practise, the better you will become.
You are doing great. Keep learning. Keep exploring. Keep building. The world of robotics is waiting for you!
End of Module Seven
Next: Module Eight β Capstone Project
Welcome to Module Eight of Level Three! This is the final module of the course. You have learned so much. You learned about kinematics, dynamics, probabilistic robotics, motion planning, robot learning, manipulation, human-robot interaction, and advanced application domains. Now you will put everything together in a Capstone Project.
A capstone project is a big project that brings together everything you have learned. It is like the roof of a house. All the walls, doors, and windows come together to make a complete building. Your capstone project is the roof that completes your learning.
In this module, you will:
This is the most exciting part of the course. You will become a real robotics engineer. You will create something new. You will solve a problem. You will show the world what you can do.
Let us begin!
By the end of this module, you will be able to:
Once upon a time, in the city of Abuja, Nigeria, there was a big Robotics Exhibition. Students from all over the country came to show their robots. The best project would win a prize.
A group of students from Enugu, led by a girl named Ada, came to the exhibition. They had worked for many months on their project. They had learned about kinematics, dynamics, probabilistic robotics, motion planning, robot learning, manipulation, human-robot interaction, and advanced application domains. Now they were ready to show what they could do.
Their project was a Hospital Delivery Robot. It would help nurses deliver medicine to patients. It would navigate hospital corridors. It would avoid obstacles. It would pick up medicine boxes and deliver them to the right rooms.
The students had worked hard. They had planned the project. They had designed the robot using CAD. They had built the chassis. They had added motors, sensors, and a small computer. They had written programs using everything they had learned.
At first, the robot did not work well. It got stuck in corridors. It dropped medicine boxes. It got confused about which room to go to.
But the students did not give up. They debugged the problems. They improved the design. They tested again and again.
Finally, the robot worked. It navigated the hospital. It avoided obstacles. It picked up medicine boxes. It delivered them to the right rooms. It was a success.
At the exhibition, the students presented their robot. The judges were impressed. The robot won first prize.
But more importantly, the students learned something powerful: with planning, hard work, and teamwork, you can build anything.
That is what you will do in this module. You will plan, design, build, program, and test your own robot. You will create a capstone project.
Let us begin!
A capstone project is a big project that brings together everything you have learned. It is the final project of a course.
A capstone project shows what you can do. It proves that you have learned the skills. It is a chance to be creative and solve a real problem.
Think of building a house. You learn to lay bricks. You learn to install pipes. You learn to wire electricity. The capstone project is building the whole house. It uses all your skills.
A university student might do a capstone project to build a robot that helps doctors.
A school student might do a capstone project to build a robot that cleans the classroom.
A hobbyist might do a capstone project to build a robot that waters plants.
A Nigerian student might do a capstone project to build a robot that helps farmers in their village.
Capstone Project
Learn Skills
|
V
Apply Skills
|
V
Build Project
|
V
Present Project
|
V
Celebrate π
A capstone project is a big project that brings together everything you have learned. It shows what you can do.
Choosing a project idea means deciding what your robot will do. It should solve a real problem.
A good project idea keeps you motivated. It makes the project meaningful. It helps you focus your learning.
Think of what problems you see around you. A problem could be: too much traffic, dirty streets, or water waste. Your robot can help solve one of these problems.
A student might choose to build a robot that sorts recycling.
A student might choose to build a robot that cleans the classroom.
A student might choose to build a robot that feeds pets.
A student might choose to build a robot that helps farmers detect ripe crops.
Choosing a Project Idea
Look around
|
V
Find a problem
|
V
Think of a robot solution
|
V
Check skills and materials
|
V
Choose idea
Choosing a project idea means deciding what your robot will do. It should solve a real problem.
Planning means thinking about what you will do before you do it. It includes setting goals, making a schedule, and listing what you need.
Without planning, projects become messy. You might forget something. You might run out of time. Planning helps you stay organised.
Think of planning a party. You decide the date. You make a guest list. You buy food. You decorate. That is planning.
A builder plans a house before building. They make a blueprint and schedule.
A student plans a science project. They decide what to research and when to submit.
A parent plans a family trip. They decide where to go, how to get there, and what to pack.
A trader plans her day. She decides what to buy, where to sell, and how much to charge.
Planning Process
Goal
|
V
List needs
|
V
Make schedule
|
V
Assign tasks
|
V
Set checkpoints
Planning means thinking before doing. It helps you stay organised and finish your project on time.
Designing means creating a plan for your robot. It includes drawing the shape, choosing materials, and deciding how parts fit together.
A good design makes building easier. It prevents mistakes. It ensures the robot works well.
Think of drawing a picture before painting it. The drawing is the design. The painting is the build.
Car designers draw cars in CAD before building them.
Students design their robot chassis before building.
A carpenter designs a chair before cutting wood.
A tailor designs a dress before sewing.
Designing a Robot
Sketch on paper
|
V
Draw in CAD
|
V
Choose materials
|
V
Place motors and sensors
|
V
Check measurements
|
V
Final design
Designing means creating a plan for your robot. It includes drawing, choosing materials, and deciding how parts fit together.
Building means putting the robot together using your design.
Building turns your design into a real robot. It is where you see if your plan works.
Think of assembling a jigsaw puzzle. You follow the picture to put the pieces together. Building a robot is like that.
A car factory builds cars on an assembly line.
Students build their robot in the lab.
A child builds a toy with LEGO blocks.
A mechanic builds a generator from parts.
Building a Robot
Chassis
|
V
Motors and wheels
|
V
Sensors
|
V
Wires
|
V
Brain
|
V
Test
Building means putting the robot together. It turns your design into a real robot.
Programming means writing instructions for your robot. It tells the robot what to do.
Without a program, the robot cannot do anything. Programming brings the robot to life.
Think of giving directions to a friend. You say: "Go straight. Turn left. Stop." That is programming.
A washing machine has a program for washing clothes.
Students program their robot to navigate a maze.
A microwave has a program for heating food.
A POS machine has a program for processing payments.
Programming a Robot
Plan
|
V
Write code
|
V
Upload
|
V
Test
|
V
Debug
|
V
Done
Programming means writing instructions for your robot. It tells the robot what to do.
Testing means running your robot to see if it works correctly.
Testing finds problems before you present your robot. It helps you fix mistakes.
Think of tasting food before serving it. If it needs salt, you add salt. Testing a robot is like tasting food.
Car manufacturers test cars before selling them.
Students test their robot in the lab.
You test a new phone before using it.
A mechanic tests a generator before delivering it.
Testing a Robot
Test part 1
|
V
Test part 2
|
V
Test whole robot
|
V
Find problems
|
V
Fix problems
|
V
Test again
Testing means running your robot to see if it works. It helps you find and fix problems.
Debugging means finding and fixing problems in your robot.
Every robot has problems at first. Debugging helps you fix them. It makes your robot work correctly.
Think of a puzzle with a missing piece. You look for the missing piece and put it in. Debugging is like finding the missing piece.
| Problem | Cause | Solution |
|---|---|---|
| Robot does not move | Loose wire or dead battery | Check connections and power |
| Robot moves wrong | Program logic error | Debug the program |
| Sensor gives wrong reading | Needs calibration | Calibrate the sensor |
| Robot wobbles | Unbalanced design | Lower center of gravity |
| Robot stops randomly | Loose connection | Check all wires |
If your TV remote does not work, you check the batteries. That is debugging.
If your robot does not move, you check the motor wires.
If your fan does not spin, you check the plug.
If your generator does not start, you check the fuel and battery.
Debugging
[ Robot not working ]
|
V
[ Check sensors ]
|
V
[ Check motors ]
|
V
[ Check program ]
|
V
[ Fix problem ]
|
V
[ Test again ]
Debugging means finding and fixing problems in your robot. It makes your robot work correctly.
Documenting means writing down what you did. It includes drawings, notes, and photos.
Documentation helps you remember what you did. It helps others understand your project. It is required for presentations.
Think of a diary. You write what you did each day. Documentation is like a diary for your project.
Scientists document their experiments in notebooks.
Students keep a project notebook.
A cook writes down a recipe.
A trader keeps a record of sales.
Documentation Project Notebook +-------------------+ | Date: 10/10/2026 | | What I did: | | - Built chassis | | - Attached motors | | Problems: | | - Wires loose | | Solutions: | | - Tightened wires | +-------------------+
Documenting means writing down what you did. It helps you remember and helps others understand your project.
Presenting means showing your project to others. It includes explaining what it does and how it works.
Presenting shows what you have learned. It helps others understand your work. It is a chance to be proud of your achievement.
Think of show-and-tell at school. You show your project and tell others about it. That is presenting.
Scientists present their research at conferences.
Students present their projects at science fairs.
You show your new toy to your friends.
A trader shows new goods to customers.
Presenting
Prepare
|
V
Practice
|
V
Show robot
|
V
Explain
|
V
Answer questions
|
V
Celebrate π
Presenting means showing your project to others. It shows what you have learned.
Working in a team means collaborating with others to complete a project.
Teamwork makes projects easier. Different people have different skills. Together, you can do more.
Think of a football team. Each player has a role. Together they win. Teamwork in robotics is the same.
| Role | Responsibility |
|---|---|
| Project Leader | Keeps the team on track |
| Designer | Draws the robot and chooses materials |
| Builder | Assembles the robot |
| Programmer | Writes the code |
| Tester | Tests the robot and finds problems |
| Documenter | Writes down what the team does |
Engineers work in teams to design cars.
Students work in groups for science projects.
A family works together to clean the house.
Workers in a factory work as a team to assemble products.
Teamwork
Leader
|
+-- Designer
|
+-- Builder
|
+-- Programmer
|
+-- Tester
|
+-- Documenter
Working in a team means collaborating with others. Different roles make the project easier.
Solving real problems means using your robot to help people in real life.
Robots are most useful when they solve real problems. They can help farmers, doctors, traders, and many others.
Think of a problem in your community. Your robot can help solve it. That is solving a real problem.
A robot that delivers medicine in hospitals.
A robot that cleans the classroom.
A robot that waters plants.
A robot that helps farmers detect ripe tomatoes.
Solving Real Problems
Problem
|
V
Think of solution
|
V
Build robot
|
V
Test
|
V
Solve problem
Solving real problems means using your robot to help people. Robots are most useful when they solve real problems.
Improving means making your project better after testing.
No project is perfect the first time. Improving makes it better. It shows you are learning.
Think of writing a story. You write it once. Then you read it and make it better. That is improving.
Car companies improve cars every year.
Students improve their projects after feedback.
You improve a recipe after tasting it.
A trader improves her stall after customer feedback.
Improving
Test
|
V
Find problems
|
V
Think of solutions
|
V
Make changes
|
V
Test again
Improving means making your project better. It shows you are learning and growing.
Celebrating means being proud of what you have done.
You have worked hard. You have learned a lot. You deserve to celebrate.
Think of finishing a race. You cross the finish line. You cheer. That is celebrating.
Graduates celebrate after finishing university.
Students celebrate after presenting projects.
Families celebrate birthdays and achievements.
Communities celebrate festivals and successes.
Celebrating
Finish project
|
V
Show to others
|
V
Take photos
|
V
Be proud
|
V
Celebrate π
Celebrating means being proud of what you have done. You have worked hard and deserve to celebrate.
What comes next means thinking about your future in robotics.
Learning never stops. There is always more to learn. Thinking about the future helps you plan your next steps.
Think of climbing a mountain. You reach one peak. Then you see another peak. You keep climbing. Learning is like that.
Engineers keep learning new skills throughout their careers.
Students continue to advanced robotics courses.
Hobbyists keep building and improving.
Nigerian students can join robotics clubs and competitions.
What Comes Next?
Finish Level 3
|
V
Learn more
|
V
Build more
|
V
Enter competitions
|
V
Become an expert
What comes next means thinking about your future. Learning never stops.
| Word | Simple Definition |
|---|---|
| Capstone Project | A big project that brings together everything you have learned. |
| Project Idea | What your robot will do. |
| Planning | Thinking before doing. |
| Designing | Creating a plan for your robot. |
| Building | Putting the robot together. |
| Programming | Writing instructions for your robot. |
| Testing | Running your robot to see if it works. |
| Debugging | Finding and fixing problems. |
| Documenting | Writing down what you did. |
| Presenting | Showing your project to others. |
| Teamwork | Working together with others. |
| Improving | Making your project better. |
| Concept | Real-Life Example |
|---|---|
| Capstone Project | A university student builds a robot for their final project. |
| Choosing an Idea | Finding a problem in your community. |
| Planning | A builder makes a blueprint and schedule. |
| Designing | Car designers draw cars in CAD. |
| Building | A car factory builds cars on an assembly line. |
| Programming | A washing machine has a program. |
| Testing | Car manufacturers test cars before selling. |
| Debugging | Checking batteries when a remote fails. |
| Documenting | Scientists write in notebooks. |
| Presenting | Scientists present at conferences. |
| Teamwork | Engineers work in teams. |
| Improving | Car companies improve cars every year. |
| Celebrating | Graduates celebrate after finishing university. |
| Concept | Nigerian Example |
|---|---|
| Capstone Project | A Nigerian student builds a robot for a national competition. |
| Choosing an Idea | A student builds a robot to help farmers detect ripe crops. |
| Planning | A trader plans her day. |
| Designing | A tailor designs a dress before sewing. |
| Building | A mechanic builds a generator. |
| Programming | A POS machine has a program for payments. |
| Testing | A mechanic tests a generator before delivering. |
| Debugging | Checking fuel and battery when a generator fails. |
| Documenting | A trader keeps a record of sales. |
| Presenting | A trader shows new goods to customers. |
| Teamwork | Factory workers assemble products as a team. |
| Improving | A trader improves her stall after feedback. |
| Celebrating | Communities celebrate festivals and successes. |
| Concept | Everyday Example |
|---|---|
| Capstone Project | Building a piece of furniture from scratch. |
| Choosing an Idea | Deciding what to cook for dinner. |
| Planning | Planning a family trip. |
| Designing | A carpenter designs a chair. |
| Building | Building a toy with LEGO. |
| Programming | Setting a microwave timer. |
| Testing | Testing a new phone. |
| Debugging | Checking the plug when a fan stops. |
| Documenting | Writing a recipe. |
| Presenting | Showing a new toy to friends. |
| Teamwork | A family cleaning the house together. |
| Improving | Improving a recipe after tasting. |
| Celebrating | Celebrating a birthday. |
| Mistake | Why It Is Wrong | How to Fix It |
|---|---|---|
| Not planning | Project becomes messy. | Make a plan first. |
| Choosing a project that is too hard | You cannot finish it. | Choose something achievable. |
| Not testing | Problems go unnoticed. | Test everything. |
| Not debugging | Robot does not work. | Fix problems immediately. |
| Not documenting | You forget what you did. | Write everything down. |
| Not practicing presentation | You get nervous. | Practice before presenting. |
+-------------------+
| Choose Idea |
+-------------------+
|
V
+-------------------+
| Plan |
+-------------------+
|
V
+-------------------+
| Design |
+-------------------+
|
V
+-------------------+
| Build |
+-------------------+
|
V
+-------------------+
| Program |
+-------------------+
|
V
+-------------------+
| Test |
+-------------------+
|
V
+-------------------+
| Debug |
+-------------------+
|
V
+-------------------+
| Improve |
+-------------------+
|
V
+-------------------+
| Present |
+-------------------+
|
V
+-------------------+
| Celebrate π |
+-------------------+
( Start )
|
V
+----------------+
| Robot fails? |
+----------------+
/ \
YES NO
/ \
V V
+---------+ +-----------+
| Find | | Done |
| problem | | |
+---------+ +-----------+
|
V
+---------+
| Fix |
| problem |
+---------+
|
V
+---------+
| Test |
| again |
+---------+
|
V
( Back to start )
| Role | Responsibility | Skills Needed |
|---|---|---|
| Project Leader | Keeps team on track | Organisation, communication |
| Designer | Draws robot, chooses materials | Creativity, CAD |
| Builder | Assembles robot | Hands-on skills |
| Programmer | Writes code | Programming, logic |
| Tester | Tests robot, finds problems | Attention to detail |
| Documenter | Writes down what team does | Writing, organisation |
Week 1: Choose idea and plan
|
V
Week 2: Design
|
V
Week 3: Build
|
V
Week 4: Program
|
V
Week 5: Test and debug
|
V
Week 6: Improve
|
V
Week 7: Document
|
V
Week 8: Present
|
V
Week 9: Celebrate π
A capstone project is a big project that brings together everything you have learned.
Choosing a project idea means deciding what your robot will do. It should solve a real problem.
Planning means thinking before doing. It helps you stay organised.
Designing means creating a plan for your robot. It includes drawing and choosing materials.
Building means putting the robot together. It turns your design into a real robot.
Programming means writing instructions for your robot. It tells the robot what to do.
Testing means running your robot to see if it works. It helps you find problems.
Debugging means finding and fixing problems. It makes your robot work correctly.
Documenting means writing down what you did. It helps you remember and helps others understand.
Presenting means showing your project to others. It shows what you have learned.
Working in a team means collaborating with others. Different roles make the project easier.
Solving real problems means using your robot to help people.
Improving means making your project better. It shows you are learning.
Celebrating means being proud of what you have done. You deserve to celebrate.
What comes next means thinking about your future. Learning never stops.
In this module, you learned how to complete a capstone project. You learned that a capstone project brings together everything you have learned.
You learned how to choose a project idea that solves a real problem. You learned how to plan your project. You learned how to design your robot. You learned how to build your robot. You learned how to program your robot. You learned how to test your robot. You learned how to debug your robot. You learned how to document your work. You learned how to present your project.
You learned about working in a team. You learned about solving real problems. You learned about improving your project. You learned about celebrating your achievement. You learned about what comes next.
Most importantly, you learned that with planning, hard work, and teamwork, you can build anything. You are now a robotics engineer.
Congratulations on completing Fundamentals of Robotics Level Three!
Match the term on the left with its definition on the right.
| Term | Definition |
|---|---|
| 1. Capstone Project | A. Creating a plan for your robot |
| 2. Planning | B. A big project that brings everything together |
| 3. Designing | C. Thinking before doing |
| 4. Building | D. Writing instructions for your robot |
| 5. Programming | E. Putting the robot together |
| 6. Testing | F. Showing your project to others |
| 7. Debugging | G. Running your robot to see if it works |
| 8. Presenting | H. Finding and fixing problems |
Answers: 1-B, 2-C, 3-A, 4-E, 5-D, 6-G, 7-H, 8-F
Title: Plan a Capstone Project
Instructions:
Example:
Problem: Farmers waste time checking crops Robot solution: Crop Monitoring Robot Plan: - Week 1: Research and design - Week 2: Build chassis - Week 3: Add sensors and motors - Week 4: Program - Week 5: Test and improve - Week 6: Present Design: - Small robot with wheels - Soil moisture sensor - Camera for crop colour - Small computer
Title: My Dream Robot
Instructions:
Example:
| Problem | Robot Solution | Sensors Needed | Motors Needed |
|---|---|---|---|
| Water waste in schools | Robot that closes taps | Water sensor, touch sensor | Servo motor |
| Crowded markets | Robot that carries goods | Ultrasonic sensor | DC motors |
| Ripe crop detection | Robot that checks crops | Colour sensor | DC motors |
Title: Build a Simple Capstone Robot
Goal: Create a small robot that solves a simple problem.
Steps:
Deliverables:
Title: Complete a Capstone Project
Instructions:
Grading Criteria:
| Criteria | Points |
|---|---|
| Project solves a real problem | 20 |
| Design is complete | 15 |
| Robot is well built | 15 |
| Program works correctly | 20 |
| Documentation is clear | 15 |
| Presentation is confident | 15 |
| Total | 100 |
Congratulations! You have completed Fundamentals of Robotics Level Three. You are now ready for advanced study or real-world practice.
In the future, you can:
To prepare for the future:
Get ready for an exciting journey into the world of professional robotics!
Congratulations! You have completed Module Eight of Fundamentals of Robotics Level Three. You have learned how to complete a capstone project.
You learned that a capstone project brings together everything you have learned. You learned how to choose a project idea that solves a real problem. You learned how to plan your project. You learned how to design your robot. You learned how to build your robot. You learned how to program your robot. You learned how to test your robot. You learned how to debug your robot. You learned how to document your work. You learned how to present your project.
You learned about working in a team. You learned about solving real problems. You learned about improving your project. You learned about celebrating your achievement. You learned about what comes next.
You also learned many examples from Nigeria, from your home, from school, and from everyday life. You learned through stories, illustrations, and activities.
Most importantly, you learned that with planning, hard work, and teamwork, you can build anything. You are now a robotics engineer.
You have completed Fundamentals of Robotics Level Three. You have learned:
You are now ready for the future. You are ready for advanced study. You are ready for real-world practice. You are ready to change the world.
Keep learning. Keep exploring. Keep building. The world of robotics is waiting for you!
End of Module Eight
End of Fundamentals of Robotics Level Three
Congratulations on completing the course!