A self-learning lane keeping algorithm

Abstract:
With the progress and development of science and technology, artificial intelligence has gradually entered people's lives. Such as: voice recognition, image recognition, intelligent driving, etc. Autonomous driving is an important application area of artificial intelligence. Some self-driving cars have been tested on public roads. However, during the driving process of the vehicle, the environment outside the vehicle is constantly changing. How to develop an autonomous driving control strategy to deal with a large number of environmental variables and conditions is still a huge challenge. The deep reinforcement learning algorithm is a mature machine self-learning algorithm, which performs well on many small game tasks. We apply deep reinforcement learning algorithms to lane keeping tasks, combined with transfer learning, set up good reward feedback, and improve training methods. Finally, we conducted simulation training and testing on the Carla autonomous driving simulator. The test results show that it is feasible to apply deep reinforcement learning to the task of autonomous driving lane keeping. Finally, we put forward some ideas for applying this method to real cars.