2020/12/06 by Branka Mirchevska, Maria Hügle, Mirchevska, Branka +7
Engineering · Social Sciences · #Artificial Intelligence in Law #Autonomous Vehicle Technology and Safety #FOS: Computer and information sciences #Machine Learning (cs.LG) #Robotics (cs.RO) #Traffic control and management
paper · pdf · doi:10.48550/arxiv.2012.03234
openalex publication_date 2020/12/06 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
Well-established optimization-based methods can guarantee an optimal\ntrajectory for a short optimization horizon, typically no longer than a few\nseconds. As a result, choosing the optimal trajectory for this short horizon\nmay still result in a sub-optimal long-term solution. At the same time, the\nresulting short-term trajectories allow for effective, comfortable and provable\nsafe maneuvers in a dynamic traffic environment. In this work, we address the\nquestion of how to ensure an optimal long-term driving strategy, while keeping\nthe benefits of classical trajectory planning. We introduce a Reinforcement\nLearning based approach that coupled with a trajectory planner, learns an\noptimal long-term decision-making strategy for driving on highways. By online\ngenerating locally optimal maneuvers as actions, we balance between the\ninfinite low-level continuous action space, and the limited flexibility of a\nfixed number of predefined standard lane-change actions. We evaluated our\nmethod on realistic scenarios in the open-source traffic simulator SUMO and\nwere able to achieve better performance than the 4 benchmark approaches we\ncompared against, including a random action selecting agent, greedy agent,\nhigh-level, discrete actions agent and an IDM-based SUMO-controlled agent.\n