Adaptive Neural Network-Based Visual Target Tracking Control for Omnidirectional Mobile Robots

To address the problem of ensuring the target is within the field of view when robot vision tracking, this paper designs an adaptive neural network-based trajectory tracking controller for the motion control of an omnidirectional mobile robot during target tracking. First, the desired orientation angle of the omnidirectional robot is designed based on the relative positions of the human and the robot, taking into account the limitations of the camera field of view. Subsequently, a kinematic model of the robot is designed considering skidding and sliding environmental disturbances, and an adaptive neural network control scheme is proposed to cope with the uncertainty of the disturbance parameters. Theoretical and experimental proof that the solution is feasible. Simulation results show that the proposed control scheme can perform the target tracking task. Finally, the feasibility and effectiveness of the control scheme are verified through practical experiments.