Visual Servo Regulation of Wheeled Mobile Robots With Simultaneous Depth Identification

In this paper, an adaptive visual servoing scheme is developed to drive a wheeled mobile robot to the desired pose, wherein the unknown depth information is identified simultaneously. Specifically, system errors are selected by measurable signals at first, then the kinematics model is obtained in polar coordinates containing the unknown feature depth. On the basis of the concurrent learning strategy, an augmented adaptive updating law is constructed for the unknown feature depth using both recorded and current data. Then, the regulation controller is designed with polar-coordinate representation to drive the mobile robot to the desired pose under the nonholonomic motion constraint. Subsequently, rigorous stability analysis is conducted by utilizing Lyapunov techniques and LaSalle's invariance principle, demonstrating that the pose regulation errors and the depth identification error can converge to zero simultaneously. The performance of the proposed visual servoing method is further validated by both simulation and experimental results.