Visual servoing of mobile robots using potential fields /

Official URL: http://risc01.sabanciuniv.edu/record=b1123087 (Table of Contents)

The popularity of autonomous mobile robots have been rapidly increasing due to their new emerging application areas, from room cleaning, tourist guidance to space explorations. However, the development of a satisfactory control algorithm that will enable the autonomous mobile robots to navigate safely especially in dynamic environments is still an open research problem. In this work, a newly proposed potential field based control method is implemented, analyzed, and improvements are suggested based on experimental results obtained from a real robot. The experimental system, planned to be the groundwork for robot soccer team, consists of an overhead global vision camera, personal computer, wireless module, and a non-holonomic mobile robot, which is supposed to navigate safely among obstacles and reach its goal point. Images of the robots operating area are acquired by the camera, color processed by the PC in real-time to retrieve the required position and orientation information of the robot, goal, and obstacles. Then, the control algorithm is applied, and resulting reference wheel velocities are sent to the robot via wireless link. Finally, robot has to drive with these reference velocities for safe navigation and goal tracking. Experimental results in the form of motion history images, graphs and movies are presented to demonstrate the successful as well as problematic aspects of the implemented algorithm and the setup.