NEC/Va. Tech Smart Intersection Tech Aims to Reduce Crashes

NEC and the Virginia Tech Smart Road apply 5G and edge computing to provide warnings before crashes.

Dan Carney, Senior Editor

June 9, 2023

4 Min Read
Smart intersection
Metamorworks, iStock / Getty Images Plus, Getty Images

NEC Corporation of America has partnered with the Virginia Tech Transportation Institute (VTTI) to develop potentially life-saving smart intersection technology that uses AI-enabled cameras to monitor the flow of traffic and pedestrians through intersections.

The partners’ goal is to provide a warning system for road users when the AI detects an imminent collision. The system can also draw conclusions from long-term data to inform improvements to the layout and maintenance of intersections, as well as the timing of traffic signals.

In this project, NEC provides its 5G networked devices, while VTTI employs the Virginia Tech Smart Road to perform tests that train NEC’s AI. During these tests, cameras and private 5G base stations are mounted on signal poles of VTTI’s private testing facility so they can transmit high-definition traffic images via private 5G to the local edge computing processors that perform real-time analysis of the images using AI.

The system identified dangerous traffic conditions and hazards, such as the approach of pedestrians or animals, to provide warnings to drivers. These tests were conducted from January to March 2023 in Blacksburg, Virginia on the Virginia Smart Roads, one of the only facilities in the world that can conduct these advanced tests.

The private 5G system not only doesn’t cause covid nor exercise magical mind control over people nearby, but it also doesn’t identify or authenticate individuals the cameras see in the intersection. The tests included software-based privacy filtering based on requirements of the Virginia Department of Transportation.

The demonstration employed the n78 frequency band used by the Citizens Broadband Radio Service (CBRS) in the U.S. The base stations introduced for private 5G networking align with O-RAN Alliance specifications and the tests were conducted under the commission of Japan’s Ministry of Internal Affairs and Communications' "Verification Tests of 5G and Video Analysis Applications Using Traffic Signals in the United States."

The cellular-V2X system can relay warnings to vehicles approaching the intersection using the manufacturers’ in-vehicle alert systems and can warn pedestrians and cyclists of a hazard by projecting red light onto the ground in the intersection. Green lights can assure them of a safe condition.

Thanks to the close proximity of the edge computing processors right at the intersection and the high speed of the 5G wireless connection, the turnaround time for the AI to identify a hazard is only half a second. This speed is critical because it takes drivers between 0.8 and 1.6 seconds to respond to a warning, VTTI Senior Technology Implementer J.P. Talledo told Design News.

"The concept of infrastructure-cooperative mobility that NEC is aiming for requires highly reliable, low-latency communications via a dedicated network to predict or detect incidents near intersections,” said Mike Mollenhauer, Director of the Division of Technology Implementation at VTTI.

NEC VTTI equipment by Jacob Levin.jpeg

NEC's Masahiko Mack Nakagawa, inspects the company's 5G installation at the Virginia Smart Road at VTTI.

“Our 5G solution is a very small lightweight antenna, lunchbox-sized,” explained Masahiko Mack Nakagawa, Vice President of Corporate Business Development Division, NEC. We can mount it on a utility pole, traffic light, or building.” VTTI provided guidance on the best locations to install NEC’s equipment.

“Using that 5G we connect cameras at intersections and edge computing units at traffic lights,” he said. “The second part is the edge component doing AI analysis to provide situational awareness. The Software predicts what will happen next based on whether a car is going straight, slowing, or turning. Once the AI system detects the risk of accident, it broadcasts a warning system.”

“At the Virginia Smart Road, we tested 166 different use cases to evaluate if that system has good enough accuracy and good enough turnaround time to notify warning before the accident happens.,” Nakagawa said. The verdict was positive, he added. Yes, the system is effective at warning road users in time to prevent or minimize collisions.

VTTI Pedestrian detection.jpeg

NEC's AI can detect a fallen pedestrian in the intersection to trigger an alert to approaching drivers.

In addition to providing real-time analysis and warnings, the system can also monitor long-term trends at intersections. “We are collecting data so that the DOT knows what is happening and how they can improve the intersection, such as by changing signal timing. We can indicate what kind of solution is needed. Maybe a smart intersection is not required.” Upgrades can be as simple as pruning nearby trees to improve sight lines. “We can answer why they are having a lot of accidents or near-miss cases,” said Nakagawa.

The next step will be to apply what was learned from the controlled environment of the Virginia Smart Road to the real world. “VTTI intends to work with NEC to apply the results of these tests to intersections on live public roads," said Mollenhauer.

 

About the Author

Dan Carney

Senior Editor, Design News

Dan’s coverage of the auto industry over three decades has taken him to the racetracks, automotive engineering centers, vehicle simulators, wind tunnels, and crash-test labs of the world.

A member of the North American Car, Truck, and Utility of the Year jury, Dan also contributes car reviews to Popular Science magazine, serves on the International Engine of the Year jury, and has judged the collegiate Formula SAE competition.

Dan is a winner of the International Motor Press Association's Ken Purdy Award for automotive writing, as well as the National Motorsports Press Association's award for magazine writing and the Washington Automotive Press Association's Golden Quill award.

He has held a Sports Car Club of America racing license since 1991, is an SCCA National race winner, two-time SCCA Runoffs competitor in Formula F, and an Old Dominion Region Driver of the Year award winner. Co-drove a Ford Focus 1.0-liter EcoBoost to 16 Federation Internationale de l’Automobile-accredited world speed records over distances from just under 1km to over 4,104km at the CERAM test circuit in Mortefontaine, France.

He was also a longtime contributor to the Society of Automotive Engineers' Automotive Engineering International magazine.

He specializes in analyzing technical developments, particularly in the areas of motorsports, efficiency, and safety.

He has been published in The New York Times, NBC News, Motor Trend, Popular Mechanics, The Washington Post, Hagerty, AutoTrader.com, Maxim, RaceCar Engineering, AutoWeek, Virginia Living, and others.

Dan has authored books on the Honda S2000 and Dodge Viper sports cars and contributed automotive content to the consumer finance book, Fight For Your Money.

He is a member and past president of the Washington Automotive Press Association and is a member of the Society of Automotive Engineers

Sign up for the Design News Daily newsletter.

You May Also Like