Visual correction of orientation error for a mobile robot

Abstract

A mobile robot operating in a real world environment rarely achieves the theoretical accuracy in positioning that is possible despite the use of precise manufacturing and high resolution encoders in drive motors. This is due to internal errors to a small degree, but mainly caused by minor variations in floor flatness. A vision system is developed and implemented for a mobile robot to correct orientation errors. Straight line ceiling features are used for reference orientation during navigation. An efficient and robust algorithm is developed, tested extensively and implemented in a mobile robot for actual navigation experiments. Preliminary experimental results show that simple visual servoing with proportional control can already achieve significant improvements.