This project was developed for MTE 380 at the University of Waterloo.
It features a custom 3-DOF Stewart platform designed to automatically detect, center, and balance a ball in real time.
The system combines mechanical design, embedded control, and computer vision into a fully integrated mechatronic solution.
Special focus was placed on real-time computing and closed-loop control using a PID controller for fast and stable balancing.
▶️ First Iteration: https://www.youtube.com/watch?v=eKMGxfKeExc- ✅ Final Presentation: https://www.youtube.com/watch?v=GQNaFhVhSu0
src/– Core control code (C++, Arduino)vision/– Image processing and ball tracking (OpenCV)cad/– Mechanical models and drawings (Fusion360, Solidworks)docs/– System documentation, control design, and test resultsREADME.md– Project overview
- OpenCV (Python/C++) – Ball detection and tracking
- PID Controller – Real-time feedback control for platform stabilization
- C++ / Arduino – Servo control and logic implementation
- MATLAB – System modeling and controller tuning
- Solidworks – Mechanical design
- I²C or Serial – Communication between vision and control units
A 2-axis PID controller (X and Y) was implemented to keep the ball centered on the platform.
The ball’s position is detected via computer vision, and the platform adjusts its tilt accordingly in real time.
- Tuned manually based on system response
- Executed on microcontroller with fixed control loop
- Achieves fast reaction with minimal overshoot
✅ Completed
All components were developed, integrated, and tested.
The platform successfully performs real-time ball tracking and balancing using PID control.
This project is licensed under the MIT License – see the LICENSE file for details.