Marble Balancer Personal Project

Project Description: Designed a marble balancer system that uses real-time computer vision and PID control to balance a marble on a platform. The marble's target position can be configured to a pre-programmed sequence or to follow a finger in a real time.

Results

Finger Tracking.mp4

Finger Tracking Mode

Finger Tracking Demonstration With Plot.mp4

Finger Tracking Computer Vision and Live Position Plot

Balancing With Disturbances.mp4

Standard Balancing Mode

Matlab Plot of Recorded Position Data

System Details

Open CV Library

OpenCV was used for real-time image processing for marble tracking. A Hough Circle Transform detected the marble's position after grayscale filtering and region-of-interest extraction, supplying feedback to the control algorithm. The system also integrates HSV-based finger tracking, live controller tuning, and serial communication with an Arduino for servo commands.

PID Control

Utilizes a PID controller to regulate the marble's position by continuously comparing the measured position to a desired position. The controller adjusts the platform's tilt angle to minimize position error. Controller gains were tuned empirically.

Page updated

Google Sites

Report abuse