Robust Neural Network-Based Trajectory Tracking Control for Mobile Vehicles

The ability of neural network-based control systems for trajectory tracking in wheeled mobile vehicles was evaluated in this study.A significant challenge often encountered is the deviation from the desired trajectory, particularly in high-speed motion.A robust control scheme, designed using the Nonlinear Auto-Regressive Moving Average-Level 2 (NARMA-L2) approach, was employed to enhance the tracking performance under dynamic conditions.The NARMA-L2 controller, a well-established technique for nonlinear systems, was utilized to improve the accuracy and robustness of trajectory tracking in the presence of external disturbances and noise.In heavyduty mobile vehicles, such as agricultural machines, maintaining straight-line motion at high speeds is particularly susceptible to external load effects and system noise.The proposed control strategy integrates several stages, including system modeling, controller design, and the training of the neural network.To optimize the parameters of a proportional-integral-derivative (PID) controller, the Particle Swarm Optimization (PSO) algorithm was applied, ensuring precise regulation of the vehicle's speed.The controller generates a reference velocity, which is fed as a signal to control the motion of the left and right wheels, enabling effective steering and trajectory adherence.Simulation results demonstrate the effectiveness of the proposed controller in mitigating the impact of disturbances and load effects.The optimization of control parameters successfully minimizes the discrepancy between the left and right wheel positions, bringing them closer to zero.The robust parameter optimization approach, which was employed to counteract the influence of external loads, can significantly improve system performance under varying conditions.