A Reinforcement Learning Based Approach for Automated Lane Change Maneuvers

Pin Wang (1) , Ching-Yao Chan (1) , Arnaud de La Fortelle (2)

1 LBNL - Lawrence Berkeley National Laboratory [Berkeley]
2 CAOR - Centre de Robotique

Lane change is a crucial vehicle maneuver which needs coordination with surrounding vehicles. Automated lane changing functions built on rule-based models may perform well under pre-defined operating conditions, but they may be prone to failure when unexpected situations are encountered. In our study, we propose a Reinforcement Learning based approach to train the vehicle agent to learn an automated lane change behavior such that it can intelligently make a lane change under diverse and even unforeseen scenarios. Particularly, we treat both state space and action space as continuous, and design a unique format of Q-function approximator to estimate the total return which is an accumulated reward over a lane changing process. Extensive simulations are conducted for training the algorithms, and the results illustrate that the Reinforcement Learning based vehicle agent is capable of learning a smooth and efficient driving policy for lane change maneuvers.

Mots clés: Reinforcement Learning; Autonomous Driving; Lane Change; Vehicle Control

Domaines: Automatique / Robotique; Traitement du signal et de l'image [eess.SP]

Pin Wang, Ching-Yao Chan, Arnaud de La Fortelle. A Reinforcement Learning Based Approach for Automated Lane Change Maneuvers. IEEE Intelligent Vehicles Symposium, Jun 2018, Chang Shu, China. ⟨hal-01980542⟩