Smart intersections improve safety through connected infrastructure

Around 50 percent of car crashes resulting in a fatality or injury in the United States happen in or near intersections, according to the Federal Highway Administration. That amounts to two million intersection-related crashes annually, the majority of which are attributed to human errors, such as speeding, following too closely, fatigue and distracted driving.

Yiheng Feng, assistant professor in the Lyles School of Civil Engineering, is Purdue’s co-principal investigator in a $9.95M U.S. Department of Transportation project designed to increase safety through “smart intersections.” The next-generation traffic control system gathers and transmits information in real time to connected and automated vehicles (CAVs).

“Currently, intersections are equipped with loop detectors that are buried beneath the pavement,” Feng said. “When a vehicle drives over the loop, it communicates to the traffic signal to change. But there’s no capability to gather information when the car is not within the detector area, that’s the limitation of the loop.

“The CAV technology enables vehicles to connect with infrastructure and communicate their speed and position. Likewise, smart intersections are equipped with advanced roadside sensors such as radar, cameras and infrared cameras to detect all the objects nearby, not only the CAVs, but also pedestrians and cyclists.”

That information can be instantaneously sent to CAVs in the vicinity, triggering onboard warnings when cars are approaching dangerous situations. The University of Michigan Transportation Research Institute is heading up the project and experimental smart intersections have been installed in downtown Ann Arbor, Michigan, where 3,000 cars were equipped as connected vehicles. It sounds like a lot, but only amounted to about 3% of the total vehicle population in the area.

“We must demonstrate the benefits of connected vehicles so consumers are incentivized to purchase them,” Feng said. “There is a marginal cost associated with installing the connectivity devices but the potential safety benefits are substantial.”

When an impatient driver is sitting behind a bus that’s stopped to unload passengers, it’s not uncommon to see the driver swerve around the bus. But what if there’s a pedestrian crossing the intersection? If the intersection has sensors to identify the pedestrian, projecting both the pedestrian’s trajectory and the vehicle’s trajectory, infrastructure can detect a potential crash and warn the driver to stop. Similarly, infrastructure could detect black ice on the road ahead, warning drivers to slow down.

Feng envisions a future where all vehicles are connected, allowing for greater data collection and improved traffic management as well as safer intersections. But what about privacy concerns?

“Gathering personal identifying information from the vehicle is forbidden,” Feng said. “The smart intersections collect data on the speed and trajectory of a vehicle, but not the car’s make or model, the license plate or anything like that. It is critical that we respect privacy when conducting this type of research.”

Drivers who use Waze, Google Maps or other GPS systems to navigate their routes already receive warnings about slowdowns in traffic, debris in the path of travel or vehicles stopped along the side of the road. Implementing smart intersection and CAV technology across the nation would provide similar hazard updates, just integrated into the operation of the vehicle, rather than through an app. Smart intersections would have other benefits, too.

“In addition to increasing safety, improving mobility is another primary goal of the project,” said Feng. “Smart intersections could reduce the time drivers spend sitting at red lights when there’s no traffic coming from the other direction.”

The Purdue team will be responsible for developing core algorithms such as sensor data fusion, traffic state estimation and signal optimization that enable intelligent traffic control, Feng explained. Team members also will work with industrial partners to plan, deploy and test the developed algorithms at the smart intersection locations.