• 2 min read
AI ties road surface data to crash risk
A University of Houston study linked pavement data and 24,000+ police crash narratives to flag road segments with higher crash risk.
Image: TechXplore
A University of Houston researcher says combining road condition data with police crash narratives can help agencies pinpoint where pavement problems may be increasing crash risk.
Lu Gao, a professor of civil and environmental engineering, used artificial intelligence to analyze large roadway datasets covering pavement structure, surface condition, roadway geometry, and crash records. A case study linked more than 24,000 police crash narratives with a pavement management dataset containing about 180,000 records.
According to Gao, the results showed strong associations between friction and texture measures and wet-pavement crash mechanisms. The findings were published in Accident Analysis and Prevention.
Recommended reading
Evolute i-Sky slips to fall 2026 launch
“A case study using over 24,000 police crash narratives linked to a pavement management dataset of approximately 180,000 data records demonstrates strong associations between friction/texture measures and wet-pavement crash mechanisms.”
The work uses large language model-based crash narrative analysis to turn unstructured police reports into structured labels such as hydroplaning or curve-related loss of control. That matters because those details are often buried in free-text narratives and missing from standard crash databases, making them difficult to extract through manual review or simple keyword matching.
Gao said the goal is to help transportation agencies identify road segments that most need maintenance or safety upgrades, so limited budgets can be aimed at places where improvements could have the biggest effect on pavement conditions and crash reduction.
The study focused on pavement measures including roughness and skid severity. Prior research, Gao said, has found that highly rough pavement can raise crash frequency, while skid resistance is strongly negatively correlated with crash occurrence, especially in wet conditions.
The paper is: Sarayu Varma Gottimukkala et al, “Integrating pavement condition records with LLM-based crash narrative analysis for pavement safety assessment,” Accident Analysis & Prevention (2026). DOI: 10.1016/j.aap.2026.108609
Frontier Editor
Dan is our resident futurist, covering electric mobility, space exploration, and the smart home. He's interested in atoms just as much as bits. Whether it's a new battery chemistry, a reusable rocket, or a protocol that finally makes IoT devices talk to each other, Dan breaks down the engineering that pushes humanity forward.
via TechXplore
