Event Details

Robust Control of Underactuated Mechanical Systems

Presenter: Chun-Yi Su - Department of Mechanical Engineering, Concordia University
Supervisor:

Date: Fri, February 8, 2002
Time: 11:00:00 - 12:00:00
Place: EOW 430

ABSTRACT

Abstract:

In recent years, there has been great theoretical and practical interest in controlling underactuated mechanical systems. These systems are defined as underactuated because they have more joints than control actuators. Much of this interest is a consequence of the importance of such systems in application. For example, underactuation may arise in free-flying space robots, underwater vehicles without base actuators, legged robots with passive joints, redundant robots with flexible components, and in many other practical applications. Furthermore, when one or more joints of a standard manipulator fail, it becomes an underactuated mechanism and needs a special control algorithm to continue operation; thus the development of a control technique for underactuated systems will increase the reliability and fault-tolerance of current and future robots. Interest in studying underactuated mechanical systems is also motivated by their role as a class of strongly nonlinear systems where complex internal dynamics, nonholonomic behavior, and lack of feedback linearizability are often exhibited. Traditional nonlinear control methods are insufficient in these cases and new approaches must be developed.

In this seminar, an entirely new method is discussed. A robust nonlinear control law is proposed for underactuated mechanical systems in the presence of parameter uncertainties. The development is based on variable structure theory. The main advantage of the presented scheme is that the uncertainty bounds, needed to design the control law and to prove globally asymptotic stability, depend only on the upper bounds of the inertia parameters. These upper bounds can easily be computed making a control law possible for complex underactuated systems. Finally, the real-time application of this algorithm to a specific underactuated robot, Pendubot, is included to demonstrate the control performance.