Design, implementation and control of a two-link fully-actuated robot capable of online identification of unknown dynamical parameters using adaptive sliding mode controller

Abstract

For years, many manufacturing and production lines use robots to accelerate the manufacture rate and present high quality products. The main objective of these robots is tracking a desired path with the least error, but generally it's impossible to overcome errors entirely due to nonlinearity of dynamic systems and time dependence of the system dynamical parameters. To overcome this obstacle, a new adaptive sliding mode controller is employed to achieve not only perfect tracking, but also to identify the exact unknown dynamical parameters' values. To investigate the ability of distinct controllers practically, a full actuated dual-link robot is provided as a test bed with direct access to manipulation torques of actuators. For this purpose, three different control techniques have been applied to the robot manipulator. First, closed loop PD controller and a robust sliding mode controller have been employed. Then, a robust model reference adaptive control (MRAC) has been developed and implemented on the robot manipulator. The results indicate that due to the system nonlinearity, only adaptive controllers can handle the robot parameters variations in both simulation and experimental test. The comparative performance of adaptive and nonadaptive controllers shows that with the same path trajectory, the adaptive controller can reduce errors significantly because of the online identification of unknown dynamical parameters of the system.