Hybrid Regressor and Approximation-Based Adaptive Control of Robotic Manipulators with Contact-Free Motion

This paper proposes a hybrid adaptive controller based on both regressor and approximation techniques. The equation of motion of the target manipulator is separated into two terms terms: (i) linear parameterization of the known modeling of the robot dynamics (inertia, Coriolis and centrifugal matrices and gravity vector) by using regressor technique, and (ii) approximation of disturbance terms (e.g. unmodeled dynamics of friction) in terms of weighting and orthogonal basis function matrices. The idea behind the paper is an estimation of unmodeled dynamics, which could be hard modeled with conventional regressor unless some modifications of control law are used. Regressor works well only when it is correct. In contrary, approximation approaches provide additional robustness. The proposed Lyapunov's stability-based control law consists of four terms: a regressor term, an approximation term, a high-gain feedback term, and a robust sliding term for compensation of modeling errors. The efficacy of the proposed controller is proved via simulation of 2-R robots.