Composite Learning Exponential Tracking Robot Control With Uncertain Kinematics and Dynamics

Abstract

For the existing adaptive robot controllers considering kinematic and dynamic uncertainties, the strict persistent excitation is necessary for parameter convergence. To alleviate this stringent constraint and improve identification and tracking capabilities, a composite learning control strategy is proposed for task space trajectory tracking. First a task space control structure with separate kinematics and dynamics is designed, then composite learning technique is introduced to parameter identification process. The asymptotical stability is proved using Lyapunov methods. Besides, the exponentially converge of kinematic and dynamic estimation error as well as the exponential trajectory tracking is guaranteed when the interval excitation condition holds. Simulation results on a planar robot model show the strategy’s effectiveness.