Autonomous Visual Control of a Mobile Robot

Abstract

An autonomous mobile robot with a vision based target acquisition system must be able to find and maintain fixation on a moving target while the system itself is in motion. This capability is achieved by most animate systems, in addition to man, but has proven to be difficult for artificial systems. We propose that efficient and extensible solutions to the target acquisition and maintenance problem may be found when the machine sensor-effector control algorithms emulate the mechanisms employed by biological systems. In nature, motion provides the foundation for visual target detection, acquisition, tracking and trailing, or pursuit. We present in this paper a summary of some simple and robust visual motion based mechanisms we have developed to solve these problems, and describe their implementation in an autonomous visually controlled mobile robot.