The Virginia Tech Transportation Institute (VTTI) and NEC, an information and communications technology leader with more than 120 years of experience, have successfully carried out a proof of concept to provide safety information and warnings to drivers, pedestrians, and other road users near intersections using artificial intelligence (AI)-based video analysis technologies.

The project, which finished its first phase in September, analyzed conditions inside and outside an intersection at the Virginia Smart Roads, using information gathered from cameras and video analysis equipment installed on traffic signal poles.

“VTTI believes this proof of concept will complement cellular-vehicle-to-everything [C-V2X] technology and improve road safety by informing vehicles and pedestrians of possible traffic hazards when passing through an intersection,” said Mike Mollenhauer, director of the division of technology implementation at VTTI. “The insitute has worked alongside NEC to develop the results of this deployment and intends to work together to apply it to intersections on live public roads. This promising achievement brings connected vehicle technology to fruition and will provide timely safety information to road users.”

NEC provided AI-based video analytics technology that detects, categorizes, and tracks road users such as vehicles, pedestrians, and bicycles around intersections. Integrating this technology with a C-V2X system notifies vehicles and pedestrians of safety information and warnings, infrastructure and vehicle information exchanges via wireless communication were evaluated.

NEC and VTTI also worked on traffic safety use cases such as predicting collisions between vehicles and pedestrians, between vehicles, pedestrian crossing detection, and right/left turn vehicle detection.

In the case of predicting collisions, vehicles were successfully notified within 0.5 seconds after the video analysis detected the possibility of a collision. This makes it possible to issue warnings earlier than a driver’s response time after a possible collision is detected. This is expected to be useful in reducing traffic accidents by quickly communicating the risk of collisions at intersections with poor visibility, etc.

“This PoC combines state-of-the-art AI-based video analytical solutions with C-V2X technology to create social value that contributes to safety and security for vehicles and pedestrians in the vicinity of intersections. These technologies are capable of identifying hazardous situations in areas that can be blind spots for drivers and pedestrians and notifying them of such situations. NEC believes that this solution can contribute to a reduction in crashes and fatalities, as well as to supporting automated driving from roadside infrastructure,” said Koji Kikuchi, senior director, Smart City Business Development Division at NEC.

These advancements in intersection safety have proven effective in reducing collision among vehicles, cyclists, pedestrians, and more. Future collaborations between NEC and VTTI will include enhancing the technologies and ensuring the effectiveness of these technologies in keeping all road users safe.

Cameras and sensors installed on a light post on the Virginia Smart Roads
Cameras and sensors installed for the system were installed on a light post on the Virginia Smart Roads. Photo by Jacob Levin for Virginia Tech.
Human-machine-interface inside of a vehicle displaying a warning that a pedestrian is ahead
Human-machine-interface displaying an alert of a pedestrian ahead. Photo courtesy of Jacob Levin for Virginia Tech.
Share this story