Kinematic redundancy may be an efficient way to improve the performance of parallel manipulators. Nevertheless, the inverse kinematic problem of this kind of manipulator presents infinite solutions. The selection of a single kinematic configuration among a set of many possible ones is denoted as redundancy resolution. While several redundancy resolution strategies have been proposed for planning the motion of redundant serial manipulators, suitable proposals for parallel manipulators are seldom. Redundancy resolution can be treated as an optimization problem that can be solved locally or globally. Gradient projection methods have been successfully employed to solve it locally. For global strategies, these methods may be computationally demanding and mathematically complex. The main objective of this work is to exploit the use of differential dynamic programing (DDP) for decreasing the computational demand and mathematical complexity of a global optimization based on the gradient projection method for redundancy resolution. The outcome of the proposed method is the optimal inputs for the active joints for a given trajectory of the end-effector considering the input limitations and different cost functions. Using the proposed method, the performance of a redundant 3PRRR manipulator is investigated numerically and experimentally. The results demonstrate the capability and versatility of the strategy.
Inexact primal–dual gradient projection methods for nonlinear optimization on convex set