Page 186 -

Chapter 4

Computed-Torque Control

In this chapter we examine some straightforward control schemes for robot
manipulators that fall under the class known as “computed-torque controllers.”
These generally perform well when the robot arm parameters are known
fairly accurately. Some connections are given with classical robot control,
and modern design techniques are provided as well. The effects of digital
implementation of robot controllers are shown. Trajectory generation is
outlined.

4.1 Introduction

A basic problem in controlling robots is to make the manipulator follow a
preplanned desired trajectory. Before the robot can do any useful work, we
must position it in the right place at the right instances. In this chapter we
discuss computed-torque control, which yields a family of easy-to-understand
control schemes that often work well in practice. These schemes involve the
decomposition of the controls design problem into an inner-loop design and
an outer-loop design.
In Section 4.4 we provide connections with classical manipulator control
schemes based on independent joint design using PID control. In Section 4.6
we show how to use some modern design techniques in conjunction with
computed-torque control. Thus this chapter could be considered as a bridge
between classical design techniques of the sort used several years ago in robot
control, and the modern design techniques in the remainder of the book which
are needed to obtain high performance in uncertain environments.
We assume here the robot is moving in free space, having no contact with
its environment. Contact results in the generation of forces. The force control

169
Copyright © 2004 by Marcel Dekker, Inc.