← All BooksKorean

Learning Physical AI with RoboCo: Robot Structure and Gait Control

Change a four-legged robot's gait and posture in RoboCo and see how walking works

QUAD-01WALKPhysical AI with RoboCo

Table of Contents

Roadmap. PARTs 1–5 run left to right, with the practice models and chapters each one uses below.
Roadmap Five PARTs, one quadruped, and the practice models each PART uses.Zoom ↗ (open larger in a new tab)

After the prologue, how to use this book, and getting started, the book runs through 5 PARTs and 25 chapters. Each chapter has three hands-on sections and a practice guide box.

Before you start

  1. What you'll accomplish
    The main mission: stand → walk → turn → walk again → stop at the goal
  2. Prologue
    What it means to walk on legs instead of wheels
  3. How to use this book
    Run the finished gait first, and change only one value at a time
  4. Getting started
    Setting up RoboCo and the first run: seven practice models and their start files

PART 1 · EXPERIENCE

Try it hands-on: the quadruped's first steps

No code yet: walk and perform using only WALK's direction keys and ACTION's number keys.

  1. CHAPTER 01. Get a four-legged robot walking from its first step
    1.1 Load the finished QUAD-01 practice robot
    1.2 Stand, walk forward, and stop
    1.3 Watch the feet move and the body follow
  2. CHAPTER 02. Three poses: stand, sit, and stretch
    2.1 Run Stand, Sit, and Stretch
    2.2 Pick the next move when a pose finishes
    2.3 Spot the differences in waving and stretching
  3. CHAPTER 03. Many ways to walk with the keys
    3.1 Learn WALK's six direction keys
    3.2 Watch auto modes 1, 2, and 3
    3.3 How do step size and speed look?
  4. CHAPTER 04. Walking forward, turning, and stopping
    4.1 Walk forward, turn, and stop
    4.2 Change the way the body faces
    4.3 Stop and pick up control again
  5. CHAPTER 05. Free play
    5.1 Pick a favorite move in WALK
    5.2 Pick a favorite performance in ACTION
    5.3 Get used to both robots' key styles

PART 2 · UNDERSTAND

Understand the structure: how twelve joints make one step

Check joint axes on one leg on a stand, then map four legs, twelve servos, mirroring, control layers, and coordinate frames.

  1. CHAPTER 06. How far can one leg move? (three degrees of freedom)
    6.1 See a three-joint leg's range
    6.2 Check joint axes and directions
    6.3 Name the joints after checking the real axes
  2. CHAPTER 07. Four leg positions and the parts connection table
    7.1 Set front left, front right, hind left, and hind right
    7.2 Check the part names of twelve ServoMotors
    7.3 Find a wiring error starting from one joint
  3. CHAPTER 08. Same angle, different motion (mirroring and direction)
    8.1 See the same angle move legs differently
    8.2 Set left and right legs as mirror images
    8.3 Fix a leg that moves the wrong way
  4. CHAPTER 09. Control layers from servo to gait
    9.1 How a ServoMotor moves a joint
    9.2 How joints make a leg pose
    9.3 How four legs make a gait pattern
  5. CHAPTER 10. Making all four legs face the same way (coordinate frames)
    10.1 Check the X, Y, Z axes and the origin
    10.2 Separate body coordinates from each leg's coordinates
    10.3 Read foot positions in one shared frame

PART 3 · CONTROL

Control values and actions: shape a pose by moving the feet

Foot targets and IK, computed versus actual pose, workspace and joint limits, and the posture of four legs and the body.

  1. CHAPTER 11. Changing foot targets
    11.1 Find FOOT_X, FOOT_Y, and FOOT_Z
    11.2 Move the foot one axis at a time
    11.3 Compare the target and the actual foot motion
  2. CHAPTER 12. Letting the robot work out its leg angles (IK)
    12.1 How leg lengths relate to joint angles
    12.2 Turn a foot target into joint angles
    12.3 Trace IK so it isn't a hidden calculation
  3. CHAPTER 13. Comparing computed angles with the actual leg pose
    13.1 Check IK inputs and outputs
    13.2 Send computed angles as ServoMotor commands
    13.3 Compare computed values with the real pose
  4. CHAPTER 14. Knowing in advance where a leg can't reach (workspace and limits)
    14.1 Find targets a leg can reach
    14.2 See what happens with an out-of-range target
    14.3 Keep the order: IK, limit check, command
  5. CHAPTER 15. Changing four legs and the body pose together
    15.1 Set all four foot targets together
    15.2 Link body height to leg length
    15.3 Make roll, pitch, and yaw poses

PART 4 · EXPERIMENT

Change conditions and compare: find a steadier walk

Change height, tilt, transition speed, foot path, stride, and gait cycle one at a time. The IMU/PID tilting platform is optional.

  1. CHAPTER 16. How steadily does it stand at different heights?
    16.1 Stand low, standard, and high
    16.2 Compare joint angles with height above the floor
    16.3 Find the steadiest pose
  2. CHAPTER 17. Comparing body tilt and turn
    17.1 Change side-to-side tilt (roll)
    17.2 Change front-to-back tilt (pitch)
    17.3 Split foot positions in a turned pose (yaw)
  3. CHAPTER 18. Comparing how fast poses change
    18.1 Build stand, sit, and stretch poses
    18.2 Change the time a pose change takes
    18.3 Compare fast changes with safe ones
  4. CHAPTER 19. Foot paths of standing legs and moving legs
    19.1 Tell standing legs from moving legs
    19.2 Build lift, carry, and set-down paths
    19.3 Compare foot lift height with landing results
  5. CHAPTER 20. Comparing stride, speed, and gait cycle
    20.1 Change only the stride
    20.2 Change only speed and gait cycle
    20.3 Compare distance, tilt, and number of falls

PART 5 · PROJECT

Finish and improve the mission: a walking mission

Gait cycle and leg order, walking forward and back, turning, and recovery, then repeat and improve the final mission.

  1. CHAPTER 21. Building the gait cycle and leg order
    21.1 Set the standing and moving order in one step
    21.2 Set the order the four legs move in
    21.3 Follow the four legs through the run log
  2. CHAPTER 22. Walking forward and backward
    22.1 Reverse the direction of the foot path
    22.2 Choose values for walking forward and back
    22.3 Connect starting and stopping
  3. CHAPTER 23. Turning left and right
    23.1 Make a left-right stride difference
    23.2 Compare body rotation and turning radius
    23.3 Sequence a straight walk followed by a turn
  4. CHAPTER 24. Getting a fallen robot walking again (errors and recovery)
    24.1 Find wiring, direction, and angle errors
    24.2 Find IK, joint-limit, and foot-path errors
    24.3 Stop safely, stand up, and run again
  5. CHAPTER 25. The final walking mission
    25.1 Stand, walk, turn, walk, and stop
    25.2 Record the completion rate over repeated runs
    25.3 Check the improvement before and after a change

To finish

  1. Epilogue
    How twelve joints became one step
  2. Appendix A Quick reference for running and connecting
  3. Appendix B Key maps by model, and rules for stop, return, and quit
  4. Appendix C IK workspace, joint limits, and exception record sheet
  5. Appendix D Experiment record sheets and the final mission checklist
  6. Appendix E Gait tuning values and a walking troubleshooting table
  7. Appendix F Joint, coordinate, and gait terms and units

The book is published in Korean; chapter titles here are translated.