Vermont tests AI technology to stop wrong-way drivers

The TAPCO wrong-way detection system is being piloted on the northbound and southbound interstate ramps at Exit 17 in Milton.

Vermont officials are testing a new artificial intelligence system designed to stop wrong-way drivers before they enter the interstate, as state police report an increase in incidents across the state.

Vermont State Police responded to 109 wrong-way drivers on interstate highways in 2024 and 117 in 2025, according to Trooper Chris Hein. In the first six months of 2026 alone, troopers have already responded to 61 wrong-way driving incidents.

“When vehicles are going the wrong direction and then you have a crash, the energy in those types of instances can be very severe, and we have seen fatal crashes over the years,” Hein said.

One incident this winter involved a wrong-way driver heading toward two Vermont state troopers. Hein said one trooper was able to maneuver and redirect the vehicle to a safe stop, avoiding a more serious crash.

Now, the Vermont Agency of Transportation is testing a new system that officials hope will provide an earlier warning to drivers.

The TAPCO wrong-way detection system is being piloted on the northbound and southbound interstate ramps at Exit 17 in Milton. The technology uses radar and artificial intelligence cameras to detect when a vehicle is traveling against the normal flow of traffic.

“When a vehicle enters the ramp the wrong way, the system activates,” said Derek Lyman, a state transportation engineer specialist.

The system triggers flashing lights, bright LEDs and high-intensity white lights to alert drivers to turn around. Cameras also help confirm whether a driver corrects course.

If the vehicle continues traveling the wrong way, the system automatically notifies Vermont State Police, who can determine whether an officer should be dispatched.

Neighboring states, including Massachusetts and Connecticut, are already using similar technology. Vermont officials said they will evaluate the Milton pilot over the next one to two years before deciding whether to expand the system statewide.

“If it’s successful, we’ll look to strategically deploy more,” Lyman said.