6Using Neural Networks to Avoid Robot Singularity

6.1 Introduction

In recent decades, robotics has attracted considerable attention in scientific research and engineering applications. Much effort has been spent to robotics, and different types of robots have been developed and investigated [6168]. Among these robots, redundant robot manipulators, possessing more degrees of freedom (DOFs) in joint space than workspace and offering increased control flexibility for complicated tasks, have played a more and more important role in numerous fields of engineering applications [64], 68, 69. For example, the problem of finite‐time stabilization and control of redundant manipulators is investigated in [69], and a controller is designed to attenuate the effects of model nonlinearity, uncertainties, and external disturbances and improve the response characteristics of the system. The forward kinematics of a redundant manipulator provides a nonlinear mapping from its joint space to its operating region in Cartesian space. This nonlinear mapping makes it difficult to directly solve the redundancy resolution problem at the angle level. Instead, in most approaches, the problem is first converted into a problem at the velocity or acceleration level, and solutions are then sought in the converted space. A popular method is to apply the pseudoinverse formulation for obtaining a general solution at the joint‐velocity level. However, this strategy suffers from an intensive computational burden ...

Get Kinematic Control of Redundant Robot Arms Using Neural Networks now with the O’Reilly learning platform.

O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.