PaveX: turning road inspections into color-coded data instead of guesswork
A Purdue spinout replaced expensive survey vehicles with a low-cost sensor kit and computer vision that scores every crack and pothole by depth, and has already mapped over 3,400 miles of Indiana roads.
I ran into PaveX while reading about how American cities decide which roads to repave, and the honest answer for a lot of them turns out to be "someone drove around and eyeballed it." Municipal pavement assessments have historically relied on specialized survey vehicles costing hundreds of thousands of dollars, or on inspectors making subjective calls from a moving car. Neither approach scales well when a public works department is responsible for thousands of miles of asphalt.
PaveX, a startup connected to Purdue University and founded by civil engineering professor Mohammad Jahanshahi, replaces that with a data collection kit that costs only a few thousand dollars and bolts onto any ordinary vehicle. Cameras and depth sensors feed a patent-pending computer vision pipeline that scores pavement condition automatically: standardized and repeatable, instead of depending on who happened to be driving that day.
The clever part is how the output gets presented. Cracks and potholes are color-coded by depth: a dark reddish brown means no real depth, blue means a defect of 30 millimeters or deeper. A public works crew can glance at a map and immediately see where the budget should go, without reading a spreadsheet or waiting on an engineer's report.
Since January 2025, the system has surveyed and assessed more than 3,400 miles of Indiana roads, and PaveX is now running pilots in California, Illinois, Michigan, and North Carolina. It builds on more than a decade of Jahanshahi's own research, which probably explains why it already works well enough for real government agencies to hand over their road networks to it.
It's a small, unglamorous problem, and that's exactly why it's a good fit for AI: nobody needs a chatbot on the highway, just a cheaper and more honest way to know which roads are actually falling apart. I like startups that pick a boring, expensive government workflow and just make it work.