Abstract
This paper describes a collision avoidance method for a mobile robot operated by history of human operators as well as by computerized fuzzy controller. In order to cope with changeable environment, a mobile robot has to provide higher planning which watch success, failure of plan and provide feedback of perception. In this paper, a mobile robot makes a judgment based on relative position and speed between a robot and obstacles, and it predicts next environment to some extent. Moreover a mobile robot tries to acquire a skill-based intelligent through layered neural network. Some results of computer simulation are shown to ascertain effectiveness of this two-stage avoidance control. Finally, we propose layered pass-planning by thinking level which considered robot dynamics.
