Fuzzy-based visual servo with path planning for a ball-plate system

Abstract

This paper presents a visual servo control scheme for a ball-plate system with a maze. The maze built on the plate forms obstacles for the ball and increases variety and complexity of its environment. The ball-plate system is a two degrees-of-freedom robotic wrist with an acrylic plate attached at the end effector. By using image processing techniques, the ball’s position is acquired by the visual feedback, which was implemented with a webcam and a personal computer. A fuzzy controller, which provides dexterity of the robotic wrist, is designed to decide the slope angles of the plate to guide the ball to a designated target spot. Using the method of distance transform, the path planning based on the current position of the ball is conducted to find the shortest path toward the target spot. Besides, a relaxed path, appearing to be more suitable for actual application, is provided by the obstacle’s expansion approach. Experimental results show that the presented control framework successfully leads the ball to pass through the maze and arrive at the target spot. The presented visual servo control scheme works effectively in both stabilization and tracking control missions.