Separation Approach to Adaptive Control for Robot Manipulators Without Joint-Space and Task-Space Velocity Measurements

Abstract

This article investigates the task-space adaptive control problem of robot manipulators under both uncertain dynamics and kinematics. The proposed algorithm employs dual observers for the dynamic and kinematic loops to estimate joint-space and task-space velocity signals, thereby eliminating numerical differentiation, mitigating sensor-noise effects, and avoiding reliance on high-precision velocity sensors, which collectively improve control accuracy. Moreover, the controller achieves the desired separation between the dynamic and kinematic loops even in the presence of time-varying control coefficients. This separation simplifies controller synthesis and stability analysis and reduces conservatism in controller-gain selection caused by loop coupling. Finally, the asymptotic convergence of the task-space tracking error is established via the Lyapunov stability theory. The effectiveness of the proposed scheme is demonstrated through comparative simulations and experiments.