A neural controller for collision-free movement of general robot manipulators

An approach is presented for the collision-free control of general robot manipulators moving among a changing set of obstacles. The controller, based on a layered, neural network architecture, adapts to the specific eye/hand and arm/body kinematics of any arbitrarily shaped robot during an initial, unsupervised training phase. After training, the robot selects a target point by ´glancing´ at it and the controller moves the end-effector into position. This approach, which combines neural and systolic computing principles, has several advantages over traditional algorithmic solutions and extends previous work on neural manipulator control. Results of a simulation are presented.<>