An Advanced Traffic Control Model for Connected and Automated Vehicles Reduced Vehicle and Pedestrian Delay by Up to 35 Percent in Simulation and Field Tests in Ann Arbor, Michigan.
Simulation Study at a Representative Urban Intersection and Field Testing at the Mcity 2.0 Testbed.
Ann Arbor, Michigan, United States
Seattle, Washington, United States
Multimodal Multiscale Urban Traffic Control in Connected and Automated Cities
Summary Information
Advances in connected and automated vehicles (CAVs) have the potential to reshape how traffic signals and road users interact. This study extended signal-vehicle coupled control (SVCC), a strategy that optimizes traffic signal control and vehicle trajectories, from a unimodal setting to a multiscale, multimodal SVCC (M2SVCC) framework that explicitly accounted for pedestrians, cyclists, and vehicles with different types of energy consumption. This framework uses real-time data to adjust signal timing. M2SVCC was evaluated using simulation experiments. SVCC was evaluated using the field test results conducted at the Mcity 2.0, a mixed-reality testing technology facility, in Ann Arbor, Michigan.
METHODOLOGY
The proposed M2SVCC integrated pedestrians and cyclists by adding their delays and safety constraints into the intersection-level signal optimization, while vehicle heterogeneity was handled through a unified power-based energy model for ICE, hybrid and EVs. M2SVCC optimized both traffic signal timing and vehicle trajectories, including speed, acceleration, and position, simultaneously. M2SVCC operated without a cycle-based signal timing structure and incorporated signal phases for both concurrent and exclusive pedestrian phasing.
Simulation analyses were conducted in an open-source traffic simulation software, using a representative single four-way intersection in downtown Seattle, Washington. M2SVCC was compared with actuated signal control under low, medium, and high pedestrian and cyclist demand levels. Results were compared based on fuel consumption, vehicle and bike wait time, vehicle and bike time loss, vehicle and bike queue length, pedestrian time loss, and right-turn conflicts.
Due to technical constraints, field tests at Mcity 2.0 focused only on the unimodal SVCC configuration at a single intersection. The real-world data from the Mcity 2.0 testing facility was integrated within the simulation model to assess the performance of SVCC.
FINDINGS
Results from the mixed-reality field tests demonstrated that under various CAV market penetration rates SVCC substantially reduced vehicle fuel consumption, wait time, time loss, and queue length compared to fixed-time and actuated signal control, while maintaining nearly identical intersection throughput.
- Simulation results showed that when compared to actuated traffic signal control, the proposed M2SVCC reduced pedestrian time loss by 26 to 35 percent across all demand levels, and reduced bike wait time and bike time loss by 35 to 47 percent and 27 to 39 percent, respectively.
- For vehicles, M2SVCC reduced fuel consumption by 28 to 40 percent, and vehicle wait time by 11 to 51 percent under low-to-medium pedestrian and bike demand.
- Under exclusive pedestrian phasing, M2SVCC reduced right turn pedestrian-vehicle conflicts to near zero, indicating a substantial safety improvement relative to actuated control.
- In the 100 percent CAV penetration rate, compared to fixed-time control, SVCC reduced fuel consumption by 28.5 percent, wait time by 60 percent, conflicts by 40 percent and time loss by 53 percent.
