Trajectory Tracking and Navigation Model for Autonomous Vehicles Using Reinforcement Learning

A potential answer to the issues of traffic accidents and congestion is presently autonomous driving. Even though it is operated without a human driver, an autonomous vehicle must emulate human driving habits. Because human drivers will be more motivated to trust autonomous driving systems, driving safety may increase as a result. A mixed trajectory scheduling and monitoring approach is used by the vehicle control system in this investigation. Firstly, traffic patterns and driving practises are modelled using the Artificial Potential Field (APF) method. Next, such APF values are used into the Model Predictive Control (MPC) design technique, which may enhance the control outputs and trajectory. This allows the controlled car to function while being influenced by traffic conditions and driving preferences by incorporating human driving patterns and preferences into the controller. Through simulation tests, the effectiveness of autonomous driving is assessed in two situations lane switching and vehicle following that represent two opposing driving philosophies among human drivers a cautious stance and an aggressive one. These results also demonstrate that the suggested algorithm is adequate to account for driving behaviours. This special controller can therefore be applied to the field of autonomous vehicle control.