A breakthrough study from Beijing Forestry University's School of Technology has designed a two-stage alert system that slashes collision risks in roadwork zones. Published in Accident Analysis and Prevention, a top journal in safety engineering, this AI-driven solution proactively analyzes driver behavior and road hazards to prevent accidents before they occur.

Compared to normal roadways, work zones exhibit unique traffic conditions, such as partial lane closures and speed limit adjustments. Under such circumstances, drivers traveling in closed lanes will inevitably face mandatory lane-changing processes, significantly increasing the risk of rear-end collisions, vehicles intruding into work zone areas, and other crashes. As a result, work zones present unique risks to both workers and road users due to its complex and dynamic nature. Currently, extensive research has been conducted on complex risk issues in work zones, particularly in the areas of crash risk assessment and analysis of risk influencing behaviors during mandatory lane-changing. This study developed a two-stage in-vehicle warning system aimed at refining driver behavior when approaching work zones and mitigating lane-changing risks. To accurately capture the thorough behavioral processes of drivers near work zones, a driving simulation experiment was conducted involving 38 participants of diverse genders and occupations. Different warning modes (baseline vs. two-stage warning), speed limits (60 km/h vs. 80 km/h), and visibility conditions (clear vs. fog) were incorporated into the experimental design. The entire behavioral performance was segmented into approaching and lane-changing processes. Generalized Estimating Equation (GEE) models were employed to analyze behavioral changes during the approaching process under different conditions, and Fuzzy C-Means (FCM) clustering algorithm was utilized for risk assessments. Influencing analysis was then applied to examine the relationship between behavioral changes during the approaching process and lane-changing risk levels. The findings reveal that, compared to the baseline group, the two-stage warning effectively reduced the average approaching speed by 2.25 %, increased the headway distance by 19.02 %, and advanced the starting point of lane-changing maneuvers by approximately 42.8 m on average, all of which contributed to a decrease in overall lane-changing risks. Under high speed limits, despite all drivers adhering to the speed limit, the phenomenon of relatively higher deceleration still persists, leading to unsafe and uncomfortable conditions. Additionally, the risk faced by drivers is significantly heightened when traveling at higher speeds in dense fog, further corroborating the necessity of the two-stage warning. The findings provide a theoretical foundation for the development of in-vehicle warning systems for work zone areas, offering practical implications for safe and efficient operations within work zones.

The study's lead author is Junyu Hang, a junior faculty member from Beijing Forestry University's School of Technology. The research was conducted under BFU's institutional affiliation, with Xiaomeng Li, a senior researcher at Queensland University of Technology's Centre for Accident Research and Road Safety, serving as corresponding author.

This research was supported by the Fundamental Research Funds for the Central Universities (BLX202332) and National Natural Science Foundation of China (72288101, 72171017).

Paper link: https://doi.org/10.1016/j.aap.2025.107991

Written by Hang Junyu
Translated and edited by Song He
Reviewed by Yu Yangyang