Adaptive Image-based Visual Servoing with Time-varying Learning Rates for Uncertain Robot Manipulators

In this work, we propose a new adaptive image-based visual servoing (IBVS) approach with a time-varying learning rate for robot manipulators with uncertain kinematics. A monocular camera in an eye-to-hand configuration is used to extract the image features of a target object, its centroid coordinates and projected area, allowing visual tracking applications. An indirect adaptive control method based on time-varying adaptation gains and projection operators is then designed to deal with camera model inaccuracies including large misalignment angles which affect the control direction. The Lyapunov stability theory is used to demonstrate the stability and convergence properties of the adaptive IBVS approach. The time-varying gain allows for fast exponential convergence of the system parameters towards a compact set under finite excitation condition. Numerical simulations are included to illustrate the effectiveness and feasibility of the proposed vision-based control methodology.