Optimal Control with Exact Linearization of Robotic Manipulators

Recently, many attractive optimal control schemes have been developed for real-time robotic applications. Due to the inherent complex physical structures, robotic control systems embody a number of hard features, such as multi-input and multi-output characteristics, nonlinearity, time-varying parameters, and highly coupled system dynamics. Furthermore, they require consideration of such issues as robustness to external disturbances, parameter uncertainty, sensor noise, payload changes and computational errors. Thus the development of a real-time optimal control scheme with high reliability, high robustness, high accuracy and acceptable computational burden becomes an urgent and important problem for research and development. This paper presents an optimal control scheme which involves a nice combination of self-adaptive control strategy, exact linearization with output decoupling, and the computational procedure for solving the robotic dynamic equations based on a compact Lagrangian formulation. It is found from the simulation studies on the PUMA 560 robot that the actual trajectories of the robot arm converge quickly to the desired trajectories under large initial errors and robotic parameter deviations.