Robot Manipulator Redundancy Resolution

Chapter 13Time-Varying Coefficient Aided MM Scheme

13.1 Introduction

The issue of optimal motion control in robotics is becoming a more and more appealing topic as robots are widely applied in scientific research, military and civilian areas, of which the operating environment is much more complicated than that in the past [205–209]. Redundant manipulators can achieve subtasks such as obstacle avoidance [210], repetitive motion planning [128], joint limits, and singularity avoidance [211–214], or optimization of various performance criteria [215], since they have more degrees of freedom (DOF) than required to perform a given end-effector primary task [92]. A fundamental issue in controlling such redundant manipulators is to design suitable redundancy-resolution approaches [92, 215]. The conventional solution to such a redundancy-resolution problem is the pseudoinverse-based formulation [211, 212].

In recent years, to avoid online matrix inversion and achieve higher computation efficiency, much effort has been made to find the solution of the redundancy-resolution problem via optimization techniques [92]. Generally speaking, such techniques can reformulate different schemes as a unified quadratic program (QP) that is subject to equality, inequality, and bound constraints. Since the QP is equivalent to a linear variational inequality (LVI) problem, it can be solved by many methods and techniques efficiently, for example, numerical algorithm 94LVI[136] and/or recurrent neural ...