Interactive trajectory planning using mixed integer quadratic programming

Motion planning for automated vehicles in mixed traffic, where automated and human-driven vehicles share the road, is a challenging task. To reduce the complexity of this task, modern planning approaches often assume that the future movement of surrounding vehicles can be predicted independently of the behavior of the automated vehicle. Separating the prediction of others from one's own planning can lead to suboptimal, overly conservative driving behavior, especially in highly interactive traffic situations. In this paper, we present an approach to planning cooperative, interaction-aware behavior based on multi-agent trajectory planning. Here, the prediction of others as well as the planning for the automated vehicle is solved jointly using mixed-integer quadratic programming. Uncertainties in the behavior of other road users are taken into account by utilizing different intention models. The applicability of the approach is demonstrated by numerical experiments for a lane change scenario in dense traffic.