Path Tracking Laws and Implementation of a Vision -Based Wheeled Mobile Robot

This paper has studied techniques for building system configuration, control architecture and implementation of a vision-based wheeled mobile robot. The completed WMR has been built with the dead-reckoning method so as to determine the vehicle’s velocity and posture by the numerical differentiation/integration over short travelling. The developed PID controllers show good transient performances, that is, velocity of right and left wheels can track the commands quickly and correctly. Moreover, the path tracking control laws have been also executed in the DSP-based controller in the WMR. The image-recognized system can obtain motion information 15 [frames/sec] by using H -S model, which is one of the well -known color detection methods. The better performance a vision system has, the more successful the control laws design. The WMR obtains its posture from the deadreckoning device together with the vision system. Finally, we integrate these subsystems and complete the operators of the whole system. This complementing wheeled mobile robot system can be thought of as a platform for testing various tracking con trol laws and signal filtering method. To solve the problem of position/orientation tracking control of the WMR, two kinematical optimal nonlinear predictive control laws are developed to manipulate the vehicle to asymptotically follow the desired trajecto ries. A Kalman filtering scheme is used to reduce the bad effect of the imagine nose, thereby improving the accuracy of pose estimation. The experimental system is composed of a wireless RS232 modem, a DSP -based controller for the wheeled mobile robot and a vision system with a host computer. A computation -effective and highperformance DSP-based controller is constructed for executing the developed sophisticated path tracking laws. Simulation and experimental results are included to illustrate the feasibil ity and effectiveness of the proposed control laws. Key-Words: Predictive control; Mobile robot; Path tracking; Nonlinear characteristics, Kalman filtering