As the demand for autonomous driving systems continues to rise, the need for proficient highway navigation becomes paramount. This study presents a comprehensive approach to training autonomous cars for proficient highway driving using deep reinforcement learning. The research focuses on critical maneuvers of switching lanes on the road. A sophisticated simulation environment is then employed for training, enabling safe and efficient iterations. The primary objective of this project is to develop a more adaptive and efficient navigation system using Deep Reinforcement Learning (DRL). Utilizing the HighwayEnv environment and Stable Baselines' Deep Q-Network (DQN) and Proximal Policy Optimization (PPO) algorithms, the project aims to create a model capable of real-time decision-making and multi-objective optimization, thereby improving the overall efficiency and safety of highway navigation. A comparative analysis has been performed on these two algorithms to know the best algorithm suitable for the task of switching lanes. The developed system demonstrates promising results in mastering complex highway scenarios, showcasing the potential for a safer and more efficient autonomous driving future.
Autonomous Car Driving: Advanced Maneuvers Training
21.09.2024
1290053 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Safe Swerve Maneuvers for Autonomous Driving
IEEE | 2020
|SAFE SWERVE MANEUVERS FOR AUTONOMOUS DRIVING
British Library Conference Proceedings | 2020
|Planning for Safe Abortable Overtaking Maneuvers in Autonomous Driving
ArXiv | 2021
|