Predictive Inverse Kinematics for Redundant Manipulators: Evaluation in Re-Planning Scenarios

In this paper, we analyze the effectiveness of a predictive redundancy resolution for constrained manipulators in case of on-line re-planning. The method is suitably modified to cope with re-planning issues, such as possible infeasible motions and position errors. Several re-planning scenarios are evaluated. Their definition is based on the smoothness of the re-planned task with respect to the current state of the robot. This allows a deep investigation of the behavior of the method under different conditions. Simulations results on a 7-degree-of-freedom KUKA LWR IV demonstrate remarkable advantages of the predictive method, both in terms of task error and redundancy exploitation.