CIM: Constraint-based Intersection Management in Mixture of Non-Connected Vehicles

While traffic control systems for connected autonomous vehicles (CAVs) are expected to provide multipurpose mobility services in the upcoming Mobility as a Service (MaaS) era, non-connected vehicles (NCVs) must also be considered when expanding MaaS service areas. Accordingly this paper proposes constraint-based intersection management (CIM) as a traffic control system that utilizes spatiotemporal constraints to ensure that CAVs can safely negotiate intersections with poor visibility, even in the presence of NCVs. To handle such mixed environments, it is important that CAVs be able to utilize both information from traffic control systems and data collected from their on-board sensors. To accomplish this, our CIM system utilizes local dynamic maps (LDMs) to obtain NCV information and then sends spatiotemporal constraint data to the CAVs via the traffic control system. The use of spatiotemporal constraint data provides robustness against uncertainty in both NCV behavior predictions and LDM sensing errors. Through the simulation experiments conducted in this study, we confirmed that the proposed method is both efficient and robust, and that it enables CAVs to negotiate traffic intersections safely despite any potential NCV predictive uncertainties and/or position measurement errors.