Alabama’s AI approach to road asset inventory and assessment

The Alabama Department of Transportation (ALDOT) maintains over 11,000 miles of state roads, including all interstate and state routes, and sees nearly 1,000 traffic fatalities on their roadways annually. Most of these deaths, 59%, occur in rural areas—with roadway and lane departure crashes accounting for 24% of all fatal crashes. As part of the state’s Strategic Highway Safety Plan (SHSP), ALDOT is targeting infrastructure-related causes of these crashes and infrastructure-based safety countermeasures to reduce the number and severity of traffic accidents. To understand how various roadway features contribute to these crash statistics, ALDOT needs to conduct accurate and consistent roadway asset inventory, condition assessments, and maintenance.

Elements like pavement markings, signage, and guardrails are critical safety countermeasures to prevent vehicles from colliding and striking dangerous roadside hazards. Without an inventory and precise location of these types of assets, their damage often went undetected until a collision occurred. This lack of foundational data prevented ALDOT from proactively managing [their] safety infrastructure, a gap noted in the state’s SHSP. To ensure roadway structural integrity and safety for travelers, as well as preserve long-term resilience of the state’s roads, ALDOT embarked on a project to evaluate a smart, data-driven approach to roadway asset management and maintenance.

Manual workflows and network scale

Traditionally, ALDOT teams manually drove lengthy roadway networks to survey assets and perform asset health assessments, which often either overlooked potential damage and/or provided subjective asset health scoring. The network scale made traditional inspection methods—such as manual patrols or lidar surveys—expensive, time-consuming, risky for personnel, and logistically impractical for network-wide assessment, as scoring was not done empirically, person to person. These methods were not only costly, laborious, and inefficient, but the resulting inaccuracies and data gap failed to ensure roadway safety, posing direct risks to motorists.

To address these shortcomings, ALDOT initiated a pilot project to test an AI-powered approach that focused primarily on identifying and assessing the guardrails along four key roadways, recognizing that damaged guardrails were critical to avoiding and decreasing the severity of lane and roadway departure-related accidents. The project surveyed key state routes east of Montgomery: AL-108, AL-110, AL-126, and AL-293. To detect the presence and length of the guardrails, pinpoint the exact location of any damage, calculate a deficiency score for each guardrail segment, and identify key safety components such as impact attenuators, ALDOT needed advanced roadway mapping, imaging, and condition assessment technology.