Researchers writing in Scientific Reports propose a new approach to roadway safety that treats traffic risk as something that can be actively managed before a dangerous conflict develops. The study introduces an Anticipatory Risk Field Dynamics framework that uses a digital twin to continuously map changing risk across a roadway network, rather than relying primarily on crashes, historical hotspots, or individual vehicle warnings. The system combines vehicle trajectories, speeds, acceleration, Time-to-Collision, Post-Encroachment Time, and other measures to identify areas where risk is beginning to accumulate. Connected automated vehicles then use an Anticipatory Swarm Intelligence algorithm to coordinate small adjustments, such as modest speed changes, lane changes, or creating gaps at merging locations. Instead of waiting for an individual vehicle to encounter an imminent collision and respond with emergency braking, the concept is designed to anticipate how conflicts may develop over the next several seconds and use multiple connected vehicles to reduce the underlying conditions producing that risk.

The simulation results suggest that relatively modest levels of connected vehicle deployment could produce meaningful network-wide safety benefits. At 20 percent connected automated vehicle penetration, the researchers reported a 42.8 percent reduction in high-risk events, defined as situations with a Time-to-Collision below 1.5 seconds, while average travel time increased just 3.5 percent. At 10 percent penetration, high-risk events declined 21.5 percent, and the benefits continued to increase at higher penetration levels. The system also performed well during a simulated two-lane closure, reducing the severity and spread of risk and accelerating the network’s recovery, while remaining effective under realistic V2X communication delays and packet loss. From a road safety perspective, the research is significant because it suggests connected vehicles could eventually serve not only as individually safer vehicles, but as active safety agents capable of improving conditions for everyone around them, including human-driven vehicles. However, the findings should be viewed as promising rather than conclusive because the study is entirely simulation-based. The authors specifically identify real-world testing, more sophisticated modeling of human driver responses, and inclusion of pedestrians and bicyclists as important next steps before the approach can be considered ready for widespread deployment.

You can review the full report here.