Prescribed-Time Trajectory Tracking Control of Wheeled Mobile Robots Using Neural Networks and Robust Control Techniques

This chapter considers a methodology for trajectory generation and prescribed-time control addressing the autonomy problems of a vehicular system. Specifically, a neural network-based hybrid framework combined with a computer vision approach is employed for lane identification and segmentation. Then, the data gathered from the previous stage is used to generate a feasible trajectory to be followed by the vehicular system. To this end, a prescribed time controller ensures the system follows the desired reference trajectory. The controller has a hybrid structure consisting of a fixed-time control and a stabilizing part, which does not depend on the initial conditions and system parameters. The former drives the trajectories to a neighborhood around the origin at a specified time t = T. Then, the controller switches to a robust controller that makes the tracking error converge to the origin in finite time. A numerical simulator of a scaled-wheeled mobile robot is utilized to assess the performance of the trajectory generation and controller design stages. The capabilities of the proposed approach are verified by numerical simulations encompassing the unperturbed and perturbed cases to demonstrate the robustness against uncertain real-world scenarios. Furthermore, through a qualitative study, the performance of the novel scheme is evaluated, demonstrating the efficiency and outstanding performance of the proposed trajectory generation and control stages.