Towards Cooperative Motion Planning for Automated Vehicles in Mixed Traffic
This work addresses the challenge of mixed traffic scenarios for automated driving systems, but it appears incremental as it builds on existing motion planning techniques.
The paper tackles the problem of enabling automated vehicles to plan motions cooperatively with human-driven vehicles by integrating predictions of other traffic participants into the planning process, resulting in a new cost functional and a sampling-based implementation evaluated in simulation.
While motion planning techniques for automated vehicles in a reactive and anticipatory manner are already widely presented, approaches to cooperative motion planning are still remaining. In this paper, we present an approach to enhance common motion planning algorithms, that allows for cooperation with human-driven vehicles. Unlike previous approaches, we integrate the prediction of other traffic participants into the motion planning, such that the influence of the ego vehicle's behavior on the other traffic participants can be taken into account. For this purpose, a new cost functional is presented, containing the cost for all relevant traffic participants in the scene. Finally, we propose a path-velocity-decomposing sampling-based implementation of our approach for selected scenarios, which is evaluated in a simulation.