A fast converging optimal technique applied to path planning of hyper-redundant manipulators

Abstract

A multi-pass sequential localized search technique to solve the problem of path planning of hyper-redundant manipulators for the shortest path in real-time in the presence of obstacles is proposed. The problem is approached from a control perspective as a shortest path optimal control problem, where the configuration space is searched for path points that optimize a cost function. This method addresses the “curse of dimensionality” of exhaustive search techniques via a multi-pass sequential localized search, and sensitivity to local minima of greedy approaches via a backtracking technique. Further, theoretical proof shows that the proposed technique converges to a global optimal (if only one exists) or a suboptimal (if many exist) solution. The algorithm is implemented on a 6-DOF PUMA 560 and a 9-DOF hyper-redundant manipulator arm and simulation results are analyzed for cost and time of execution.