Decentralized controllers design for nonlinear uncertain systems with application on robotic systems

This paper presents investigation into the development of decentralized sliding mode control with application to trajectory tracking for hydraulically driven revolute robot manipulators. The control of hydraulically actuated robot manipulator is very challenging due to the highly nonlinearities in its dynamics, uncertainties parameters, and variations on payload. To overcome these problems, an integrated mathematical model of an N degree-of-freedom (dof) hydraulic robot manipulator is treated as a large-scale uncertain system models with bounded uncertainties where the bounds are known. This is then decomposed into interconnected uncertain subsystems in order to apply the decentralized tracking control strategy. Sliding mode control (SMC) and proportional-integral sliding mode control strategies will be utilized to overcome the inherent nonlinear dynamics under the decentralized frameworks. These approaches were adopted to ensure the stability of the system dynamics during the sliding mode and the insensitivity to the parameter variations and disturbances. The performance and robustness of the controllers were evaluated on a 3 dof hydraulically actuated manipulator through computer simulation. The results prove that the controllers have succeeded in forcing the 3 DOF hydraulic robot manipulators to track the predefined desired trajectory at all time.

I. Introduction

Hydraulic robots are widely used in construction and mining industries. However, most of earlier work in the design of the control laws for manipulators deal with electrically actuated manipulators. In terms of hydraulic actuators, relatively less work has been done [1]. Besides, the hydraulic robot manipulator dynamics are more complex than the electrical robot manipulator dynamics, hence the modeling and control of the system is a challenging task [2]. Computer Torque Control (CTC) and PID strategies have been implemented in [3] for hydraulic robot and the performance of both methods did not give satisfactory results. The constraint with PID controller is that they are not suitable for high speed robots and in situations that require precise trajectory tracking. For the CTC, it only can be implemented to robots that allow joint torque control. Besides, the setback with CTC is, it is essentially based on exact robot arm dynamic model; the explicit use of an incorrect robot model will influence the performance of the control.