This paper describes the motion control of hyper redundant robots with a learning control scheme based on linear combination of error history. For a hyper redundant serial manipulator to achieve the motion of its end effector as a main task with a number of additional motions of other links as subtasks, several effective methods to set subtasks are proposed. The objectives of these methods are to reduce the interference between main task and subtasks, and to prevent from partially singular configuration of the manipulator. The backward learning scheme is also proposed to obtain the optimum initial configuration. Several simulations and experiments with a planar 10 R serial manipulator reveal that the proposed methods are effective and useful.
Read full abstract