Scenario-Based Decision-Making, Planning and Control for Interaction-Aware Autonomous Driving on Highways

This paper proposes an architecture for integrated decision-making, motion planning, and control in autonomous highway driving. The approach anticipates, to some degree, interactions between traffic participants and their reactive behavior to the actions of the autonomous vehicle (AV). To this end, we utilize an interaction-aware traffic prediction model to identify likely scenarios resulting from the current traffic scene, depending on the AV's tactical decision options, which are evaluated by an ensemble of Scenario-based Model Predictive Controllers to decide on lane-changing maneuvers. We conduct a validation of two versions of the scenario generation using traffic data and demonstrate the combined architecture in a simulation study.