Minimizing Traffic Congestion Through Continuous-Time Route Reservations With Travel Time Predictions

Road traffic congestion occurs as demand exceeds infrastructure's capacity. In this paper, we propose a continuous-time route reservation architecture to efficiently manage vehicle routing decisions thus to eliminate congested conditions. The related problem is formulated as a mixed integer linear programme (MILP) that routes vehicles while avoiding road segments that reach their critical density. To reduce the high computational complexity of the MILP approach, we develop a heuristic algorithm, which is shown to perform well compared to the MILP formulation. The heuristic algorithm is used to conduct extensive microsimulations across a real network topology of the San Francisco area. Finally, an extension of the proposed architecture is also developed, where the modeling uncertainties are considered through a time-varying regression method making able in real-time accurate travel-time predictions minimizing the reservation errors.