Computed torque scheme based adaptive tracking for robot manipulators

A computed torque scheme based adaptive tracking control approach for robot manipulators, which can be regarded as a modified version of Craig's adaptive control law (1987), is presented in this paper. Utilising a two component structure, this controller consists of a linear PD control component and an adaptive component. The PD control, using some reasonable estimates of system parameters, is used to stabilise the robot system in the sense that all internal signals remain bounded. The adaptive control component is to reduce the tracking errors for the resultant error dynamics. In this scheme, even though the acceleration signals are still needed in parameter estimation, there is no need to compute the inverse of the estimated inertial matrix, which is required in Craig's scheme. The adaptive controller design is based on the standard Lyapunov method and the linear-in-parameter property of robot motion equations. The scheme ensures the global convergence of the tracking errors and will approach the computed torque scheme provided that the parameter estimates converge to their true values.