Reducing Congestion by Using Integrated Corridor Management Technology to Divert Vehicles to Park-and-Ride Facilities

Research Team: Mohammad Abdullah Al Faruque (lead), Mohanad Odema, Mohamad Fakih, and Tyler Zhang

UC Campus(es): UC Irvine

Problem Statement: Considerable advancements have been made in traffic management strategies over the past few decades to enhance user mobility on freeways. Still, the continuous growth of metropolitan regions and increasing mobility needs tend to impede such progress. Pre-pandemic, the recent Urban Mobility Report showed that the total cost of traffic delay in the top urban areas in the U.S.has grown by almost 48 percent over the past decade. In many of these regions (such as in California), freeways experience a great deal of traffic congestion, arising from bottlenecks at which high-volume, free-flowing traffic transforms into tightly packed clusters of low-speed vehicles. As such, transportation planners have given special attention to the notion of integrated corridor management (ICM), which encourages the adoption of global traffic management strategies. In practice, this involves consolidating the various traffic components deployed along the corridor (e.g., ramp meter controllers) into a single interconnected system with a global view of traffic condition along the entire corridor, allowing upstream traffic components to be tuned to relieve downstream bottlenecks. In other words, an ICM strategy can coordinate various traffic control units to optimize their operations along the entire freeway, rather than just on pre-specified settings or in localized areas.

Project Description: This project presents a novel approach using ICM technology to divert connected vehicles to underutilized park-and-ride facilities where drivers can park their vehicle and access public transportation. Using vehicle-to-infrastructure (V2I) communication protocols, the system collects data on downstream traffic and sends messages regarding available park-and-ride options to upstream traffic. A deep reinforcement learning program controls the messaging, with the objective of maximizing traffic throughput and minimizing CO2 emissions and travel time. The ICM strategy is simulated on a realistic model of Interstate 5 using Veins simulation software. The results show marginal improvement in throughput, freeway travel time, and CO2 emissions, but increased travel delay for drivers choosing to divert to a park-and-ride facility to take public transportation for a portion of their travel.

Status: Completed

Budget: $80,000

Policy Brief(s):