Trajectory Tracking and Nonparametric Identification of Flexible Space Robot Manipulators

Abstract

The paper compares control strategies for four flexible space robot manipulators including the classical Slotine and Li algorithm, a simple proportional derivative controller, a singular perturbation-based controller and a nonlinear backstepping controller. All strategies are tested and compared in simulations involving endpoint target positioning while tracking a square trajectory by a two-link flexible joint space robot. Simulation results indicate that controlling both nonlinearities and joint flexibility effects improve the closed-loop behavior of the space robot where the control of nonlinearities is of greater importance. The best performance is achieved by the nonlinear backstepping control strategy. We also propose Hammerstein nonlinear joint model and introduce its nonparametric nearest neighbor kernel identification algorithm and study its convergence.