Modeling Study Found That a Cumulative-Anticipative Car-Following System Increased Overall Road Capacity by 35 Percent at 90 Percent Market Penetration.
The Impacts of a Cumulative-Anticipative Car-Following System Were Evaluated in a 3-mile Long Simulation Network.
Nationwide, United States
Safety and Mobility Evaluation of Cumulative-Anticipative Car-Following Model for Connected Autonomous Vehicles
Summary Information
The advancement of automated vehicles (AVs) and connected and automated vehicles (CAVs) presents an important opportunity to improve safety, mobility, and economic efficiency, potentially saving trillions of dollars worldwide each year. This study examined the performance of a cumulative-anticipative car-following (CACF) model, which uses Vehicle-to-Everything (V2X) communication to help CAVs anticipate the behavior of multiple nearby vehicles. Unlike traditional cooperative adaptive cruise control (CACC) systems that rely mainly on vehicle-to-vehicle (V2V) data, the CACF model incorporates both V2V and vehicle-to-infrastructure (V2I) inputs to enhance coordination and situational awareness. The study systematically compared the CACF model’s performance in terms of mobility and safety outcomes against established models, aiming to evaluate its potential in reducing traffic disturbances and mitigating bottlenecks.
METHODOLOGY
The study implemented CAV modeling after extending a default psychophysical car-following model by integrating two external driver models: the CACC model and the CACF model. A custom lane-changing algorithm which considered both cautious and aggressive driving behaviors using different safe distance parameters was designed to handle lateral vehicle control, particularly in the CACF model. The study conducted seven sensitivity analyses to evaluate the mobility and safety impacts of the CACF model, comparing it against the CACC model and the software default model. The tests included assessments of maximum throughput, performance under crash and no-crash conditions, the influence of acceleration coefficients, the effect of V2I communication range, the impact of communication signal lag, and the safety performance of the CACF model under multi-lane conditions. Simulations were conducted using a 5 km (about 3 miles) freeway network, and parameters such as vehicle input rates, market penetration of CAVs, and time-to-collision thresholds.
FINDINGS
- In the CACF model without crashes, at a 90 percent CAV market penetration rate increased the overall road capacity by 35 percent, increasing the throughput from 2350 pc/h/ln to 3350 pc/h/ln.
- Starting at 30 percent CAV penetration rate, the average travel time begins to show improvement. The average travel time at 90 percent CAV penetration rate showed a 6.6 percent and 9.1 percent improvement for the CACC and CACF models, respectively.
- Under scenarios with crashes, CACF model had a smaller number of rear-end conflicts than CACC model for 30 percent or greater market penetration rates.
