Ball Balancing on a Stewart Platform using Fuzzy Supervisory PID Visual Servo Control

Abstract

In this paper, fuzzy supervisor of PID controller for ball balancing on a Stewart platform (ball and plate system) was implemented. Small-size Stewart platform was built and used for demonstrating the application of the controller. HD web camera was the visual sensor for sensing the ball position. After image processing, by using color threshold and object detection, the ball position was a feedback signal of the controller. In ball balancing control problem, the x-direction and y-direction can be decentralized and are separated into two PID control loops. Conventional PID control might give poor performance due to nonlinearity such as friction and error from the visual tracking. Model-based control design is complicated and could be suffered from modeling uncertainties. Fuzzy controller, which is a nonlinear control, is chosen to resolve the problem. By the way, the PID controller can be enhanced by using a fuzzy supervisory system. By using a few human intuitive rules, from ball speed and ball location, control performance can be improved. Each PID controller gains were adjusted by the fuzzy supervisor. The controller generated the pitch and roll references. The reference signals were transformed to six servo motor set-points by using the inverse kinematics of the platform. PID position controllers were used for the six servo motors. The ball position results showed enhancement by the proposed control system.