Skip to Main Content
In this paper, we present a congestion-aware route planning system. First we learn the congestion model based on real data from a fleet of taxis and loop detectors. Using the learned street-level congestion model, we develop a congestion-aware traffic planning system that operates in one of two modes: (1) to achieve the social optimum with respect to travel time over all the drivers in the system or (2) to optimize individual travel times. We evaluate the performance of this system using 10,000+ taxis trips and show that on average our approach improves the total travel time by 15%.