Neural Network Optimized Controller for Motion and Position Control in Autonomous Systems
Abstract: Motion and position control are key components of autonomous navigation systems. Effective operation of Wheeled mobile robots, Autonomous Cars, Autonomous Surface Vehicles and Unmanned Combat Air Vehicles demand audacious trajectory and posture controls. In this work, a novel framework of neural network optimized controller is proposed for motion and position control of autonomous systems. Two case studies are presented to demonstrate the effectiveness of the proposed scheme. In the first case, a Lyapunov-based nonlinear controller for mobile robot trajectory tracking is optimized using the neural network. In the second case, the neural network-based optimization is performed on a Linear Quadratic Regulator-based optimal controller for a 2-Degree-of-Freedom Helicopter position control. Simulation results show that the proposed scheme achieves a faster transient response, as well as better overall error minimization, as compared to traditional control schemes. Furthermore, the proposed optimization is performed on the samples from test navigation and it does not require accurate models of the plant. Therefore, it can be used to compensate for poor control performance due to inaccurate models.
Samuel Oludare Bamgbose*, Xiangfang Li, Lijun Qian
Judging Forms – Official judges only