Robot Manipulator Redundancy Resolution

Chapter 5Redundancy Resolution via QP Approach and Unification

5.1 Introduction

A redundant manipulator is defined when more degrees of freedom (DOF) are available than the minimum number of DOF required to execute a given end-effector primary task [61, 62]. Our human arm, elephant trunk, and snake are also such redundant systems [63, 64]. Compared to non-redundant manipulators, the redundant manipulator naturally has wider operational space and extra degrees to meet more functional constraints, such as the online avoidance of joint physical limits [36] and environmental obstacles [65, 66]. One of the most fundamental issues in operating the redundant manipulators is the redundancy-resolution problem. That is, given the Cartesian velocity/acceleration trajectories of the end-effector, we are required to generate the corresponding joint velocity, acceleration, and/or torque trajectories in real time [61].

Redundancy resolution for optimizing the joint torques is one most important part of redundancy-resolution problems, which is aimed at making a more effective utilization of input power from actuators by exploiting the extra DOF in redundant manipulators. Similar to the velocity-level redundancy resolution, most researchers use the pseudoinverse-type solution for the torque-minimizing redundancy resolution of manipulators. In particular, initial formulation for local torque minimization was presented in the mid 1980s by Hollerback and Suh [67]. They designed several schemes ...