Netradyne, a computer vision startup for trucking fleets, has announced that its Driveri vision-based safety program has captured and analyzed 1 billion minutes and 500 million miles of driver video, improving driver behavior while advancing autonomous vehicle development.
“What we are doing is very very comprehensive,” Netradyne CEO Avneesh Agrawal told FreightWaves. He referred to the program, rooted in deep learning models and computer vision algorithms, as “practically the perception side of autonomous driving,” analyzing road conditions, driving events and violations during “every second of every driving day” for thousands of fleet drivers across the country.
Founded in 2015, with offices in San Diego and Bangalore, Netradyne claims as customers hundreds of commercial fleets, including rideshare, limousine service, vans and pickup trucks, utility vehicles, last-mile delivery, freight, and heavy-duty trucking.
Driveri works by uploading data collected from dashcams to the cloud, and then, based on the fleet manager’s involvement, passing on that information to the driver in the form of an app, giving the driver a report on their safety performance.
Most legacy and incumbent video safety systems are reactive, according to Agrawal, analyzing data only if a driver does something wrong.
Utilizing a driver score called a “Green Zone,” Driveri takes a different approach, capturing positive as well as negative driver responses and combining supportive comments along with constructive feedback about risky driving behavior. We want to “change the conversation” with drivers, Agrawal said. “We want to recognize the good things.”
Another application involves mining Netradyne’s millions of miles of road data to advance the autonomous driving industry, Agrawal said.
Basically, the company takes the data and images generated by Driveri, and turns them into high-definition maps of the U.S. Autonomous vehicles need to know if there is a change in the road, be it a construction zone or an accident, Agrawal observed. One way to solve that problem is to create maps that tell you every time something changes, a capability that hinges on access to massive data sets.
“The more miles you have, the more data you have, the more you can respond to changes,” said Agrawal, who believes Netradyne will play “an integral” role providing data sets for the self-driving industry.
To date, Netradyne has announced one such partnership, with Hyundai’s navigation and mapping subsidiary Hyundai MNSoft, aimed at improving the OEM’s high-definition mapping program for autonomous cars.
Agrawal says he is bullish on Netradyne’s business prospects, but what also excites him are the safety and compliance benefits. For example, the company has seen a 60% improvement in stop sign compliance across all fleets using Driveri, he said.
“The platform,” he said, “is going to save lives.”