Simulation-based reinforcement learning for autonomous driving

Christopher Galias, Adam Jakubowski, Henryk Michalewski, Błażej Osiński, Paweł Zięcina

Keywords: autonomous driving, reinforcement learning, synthetic data, domain randomization

Abstract: We use synthetic data and a reinforcement learning algorithm to train a driving policy intended to control steering of a full-size real-world vehicle in a number of restricted driving scenarios. The driving policy uses RGB images as input.