Town adopts AI road‑scoring tool; overall network score 75.2 and $1.75M estimated to maintain it

Consultants presented the results of a new pavement‑management program that uses image‑based AI to evaluate road distress every ~20 feet and roll results up to a 0–100 rating for segments and whole roads. The system covered the town's ~146 paved miles (town‑maintained) and produced an overall weighted score of 75.2.

The consultant, Tim Garrow of Beta, explained the methodology: vehicle‑mounted cameras capture images approximately every 20 feet; images are converted to still frames and analyzed by machine learning to identify cracking, potholes, alligator cracking and other distresses. Segment‑level scores are combined into road‑level averages that feed a GIS platform for planning and public reporting.

The presentation included planning‑level cost estimates showing that about 12 miles of roads are candidates for major rehabilitation (estimated at planning level at roughly $7 million), nearly 10 miles for minor rehab (roughly $5 million), about 50 miles suited for preventative maintenance and an estimated annual funding need of approximately $1.75 million to maintain the current network score of 75.2. The consultant emphasized these are planning estimates and that the tool allows the town to prioritize limited resources and avoid repaving roads that will soon be affected by other underground work.

Outcome: Council accepted the presentation and directed staff to post appropriate public materials and to use the tool for multi‑year planning and sequencing work with other capital projects (for example, timing road work around school construction).