A self-balancing robot that stays upright on a single ball. Three omniwheels, three axes of control, one impossible-looking trick. You build the hardware, you tune the controller, you understand the math.
Real control theory. Real hardware. Not a simulation.
Each step teaches something real about control systems.
Assemble the frame, mount the omniwheels, wire the motors and IMU. Solder, screw, test — your hands, your robot.
Read the 6-DOF IMU, filter the noise, fuse accelerometer and gyroscope data. Understand complementary and Kalman filters.
Get one axis working first. PID tuning, step response, overshoot. Feel the math become physical.
All three omniwheels working together. Inverse kinematics from tilt error to wheel speeds. The real challenge.
Replace PID with a full state-feedback LQR controller. Model the system, compute the gain matrix, achieve optimal balance.
Once balanced, make it move. Translate, rotate, follow paths — all while staying upright on a single ball.
The bridge between textbook theory and real engineering.
PID, sensor fusion, state-space models, LQR — implemented on real hardware, not just MATLAB. Your code, your robot.
Hardware designs, firmware, and documentation — all open. Fork it, modify it, improve it. No black boxes.
Designed for mechatronics, EE, and CS students. Covers topics from 2nd-year control systems through graduate-level optimal control.
Wi-Fi enabled, Arduino-compatible, powerful enough for real-time control loops. Tune parameters wirelessly, log data live.
Print your own frame or use the included parts. Modify the design, add sensors, make it yours. STL files included.
Assembled in Hong Kong, shipped globally. Every component included — motors, wheels, electronics, ball, fasteners. Nothing extra to buy.
Everything for a complete ballbot build.
Get notified when the OrbBot kit is available, or join a guided workshop at CreatorPort HK.