Super Human Performance in Gran Turismo Sport Using Deep Reinforcement Learning

Autonomous car racing raises fundamental robotics challenges such as planning minimumtime trajectories under uncertain dynamics and controlling the car at its friction limits. In this project, we consider the task of autonomous car racing in the topselling car racing game Gran Turismo Sport. Gran Turismo Sport is known for its detailed physics simulation of various cars and tracks. Our approach makes use of maximumentropy deep reinforcement learning and a new reward design to train a sensorimotor policy to complete a given race track as fast as possible. We evaluate our approach in three different time trial settings with different cars and tracks. Our results show that the obtained controllers not only beat the builtin nonplayer character of Gran Turismo Sport but also outperform the fastest known times in a dataset of personal best lap times of over 50, 000 human drivers. Reference: F. Fuchs, Y. Song, E. Kaufmann, D. Scaramuzza, P. Duerr SuperHuman Performance in Gran Turismo Sport Using Deep Reinforc