Artificial Intelligence

How Gainesville city officials are using artificial intelligence to improve its road conditions


Data-driven developer RoadBiotics developed the road-rating software, and it’s been implemented by municipalities in 34 states and 14 countries. 

Public works director Chris Rotalsky said the software provides an algorithm that takes pictures of roadways every 10 feet, and rates the road by segments and points.

“Utilizing this system provides several benefits to the Public Works Department,” said Rotalsky. “Staff time for data input is reduced, and the city’s street network is evaluated within a short timeframe while applying the same visual criteria for analysis to each segment.” 

According to RoadBiotics representatives, it’s all done through labeled image data. Road images are captured from a car’s windshield camera, and through-machine learning and digital paint brushing, the AI begins to scan the roads.

What does the AI look for when assessing each road image? The machine is designed to check for everything from unsealed cracks to cold patches to potholes before determining a final grade.

The system grades city roads on a five-level scale from green to red.
Dark green roads are optimal and in the best condition. Yellow and orange roads, graded between 2, 3 and 4 respectively, are declining road conditions. And the worst-conditioned roads are coded red and are rated a 5. 

According to the Pittsburgh-based company representatives, road ratings and reports are posted on an interactive, GIS-based platform called RoadWay.

Those ratings and reports help the city prioritize which roads need immediate improvement.

“The data provided from RoadBotics is combined with other rating criteria such as base condition, ride condition to help determine repair and resurfacing actions needed for the street network,” said Rotalsky.

In 2018, the city did its first road assessment using the software, with most of the city’s roads graded as green or yellow.



READ SOURCE

This website uses cookies. By continuing to use this site, you accept our use of cookies.