VARIABLE STRUCTURE CONTROLLER DESIGN FOR ROBOT MANIPULATORS

Abstract

Robots have played a dominant role in the recent trends towards automation during the past decade. The applications of robots will be virtually unlimited in the future inview of their flexibilities. However, present applications are limited to simple slow-speed, low-precision tasks. These problems are mainly due to the difficulties encountered in controlling the robot at high operational speeds. The robot manipulator is a mu Itivariable, highly non linear and dynamically coupled system. The amount of interaction between linkages is a function of the speed of the manipulator and largely responsible in complicating the control problem. However, the future use of such systems will require high speed, high precision and high accuracy. In order to improve the performance and range of operation of the manipulator we must use a dynamic control method which attends to the nonlinearity and strong coupling problems inherent with the robotic manipulator system effectively. For an effective implementation of an advanced control method for the robot manipulator we propose Variable Structure Control (VSC) strategy as a viable alternative of constraints involved in the design of conventional control methods of robot manipulators. Variable Structure System is expected to be powerful and potential tool to construct new control strategy for robot manipolators. The objective of the present research work is to obtain new solutions to the main problems that have constrained the effective implementation of controllers for robot manipulators

using the theory of Variable Structure Systems. These solutions require mathematical theories to be developed, effective control algorithms and simulation studies done that support the implementation of control laws. Necessary analysis have also been undertaken to evaluate the factors which affect the performance of Variable Structure Controllers for robot manipu lators. The development of control algorithms in the present research work is heavily influenced by the Regulated Derivative Control (RDC) strategy proposed by R.G. Morgan et al. [55]. Two new decentralized Variable Structure Control algorithms are developed to the design of controllers for robot manipulators. In the first algorithm which makes the regulation of derivative of switching variable (s), a state dependent problem. This removes some of the drawbacks and assumptions present in the RDC algorithm. A shorter reaching phase is obtained which is very much desirable in the design of Variable Structure Control System. The state trajectory of the system enters into sliding mode with the low value of controller gain. This removes some of the drawbacks like extreme sensibility of unmodelled dynamics, saturation of actuators and unnecessary stresses to the system hardware. Time of motioncurves based on the new algorithm clearly indicate the improvements. An attempt is made to remove the unfavourable criticism in the selection of weighting factors K.-| and K.j. The "cha'tter ing", which represents the imperfection in the sliding mode and may excite high frequency dynamics neglected in the course of modeling, is reduced to the lowest va1ue.

Another important contribution made in the present research is by developing a second control algorithm in the framework of VSS. The application is extended to the design controller for robot manipulators. The regulation of s consists of two terms. The first term is proportional to s only, which plays the role to ensure the equivalent system strict passive and, the second term is a traditional VSS term which plays a role to overcome system uncertainities and disturbances. The analysis reveals some \/ery important facts. The controller gain associated with the linear term is not arbitrary but related directly with the traditional VSS controller gain and the initial states to satisfy the control requirements. This lead's to generate track characteristics which provide necessary informations about reaching time. New informations are obtained about different tracks necessary to speed up the motion in the reacing phase with respect to s = 0. To obtain comprehensive informations about the performance improvements potential of the proposed control algorithms two different phases of control motions are suggested to achieve the best control objectives. The process of implementation are categorized as Reaching Phase Control (RPC) and Chattering Reduction Phase Control (CRPC). The new decentralized variable structure control strategies developed for this thesis are novel, attractive and provide solutions of many constraints, complexities and ambiguities involved in the control of robotic manipulators in the framework of variable structure control.