Learning Control of Hyper Redundant Robots Based on Linear Learning Control Scheme

摘要

This paper describes motion control of hyper redundant robots with learning control based on linear learning control scheme. In the conventional linear learning control scheme for non-redundant robots, the increment of joint inputs in a certain learning process could be calculated with an inverse Jacobian matrix and the linear combination of error vectors. By using general solution of inverse kinematics with pseudo inverse of Jacobian matrix, the control scheme was extended so that a redundant robot can realize a motion as a sub task while generating the trajectory specified as a main task. The new control scheme was applied to control a planar serial manipulator with 5 DOF, which generated continuous path trajectory while avoiding obstacles in its workspace. Several simulations and experiments revealed that the proposed control scheme was effective and useful.