This paper aims to propose an active disturbance rejection control (ADRC)-based visual servoing strategy for regulating a wheeled mobile robot from varying initial poses to a desired pose at an exponential rate. It addresses challenges associated with non-holonomic constraints, uncertain depth information and unknown translational parameters in monocular vision systems.
The uncertain depth information in monocular vision and unknown camera-to-robot translational parameters are modeled as internal uncertainties of the visual servo system. An input-state scaling technique is used to decouple the system into two subsystems, controlled by angular and linear velocities, respectively. The angular velocity controller is designed to ensure strict exponential convergence, while the internal parametric and bounded uncertainties of the system are estimated and compensated for by an extended state observer and a switching linear velocity controller.
The separate design of the angular and linear velocity controllers effectively overcomes the non-holonomic constraints of the mobile robot, ensuring robust performance under diverse conditions. Furthermore, the ADRC-based strategy successfully handles uncertain depth information and unknown translational parameters. The convergence of the error system is rigorously proven using Lyapunov theory, and simulation results verify the effectiveness of the proposed scheme.
To the best of the authors’ knowledge, this study introduces, for the first time, a novel approach that combines ADRC with visual servoing for non-holonomic mobile robots. This approach enhances the adaptability and accuracy of the robot’s navigation in environments characterized by unknown system uncertainties. The proposed method demonstrates enhanced practical performance over conventional techniques by effectively managing the inherent uncertainties of the system.
