Collision avoidance for a visuo-motor system with a redundant manipulator using a self-organizing visuo-motor map

Collision avoidance for a visuo-motor system with a redundant manipulator is described. A self-organizing map is used to resolve the inverse kinematics of the manipulator. The map determines the joint angles of a redundant manipulator so that the end effector of the manipulator reaches a target given in the image space. In our past study, although the map can make the manipulator take obstacle free poses, it is not considered if the trajectory of the manipulator avoids collisions. In this paper, we added a path planning system to the visuo-motor system. By combining the path planning and the self-organizing map, the system accomplished collision avoidance. Simulation and experimental results showed that the self-organizing map learned mapping, and that by using the map the manipulator can avoid collisions