Virginia Tech, partners design blueprint to improve self-driving vehicles with $7.5 million grant
On Oct. 11, researchers from Virginia Tech’s Transportation Institute and their partners put an autonomous Ford F-150 through a series of driving scenarios related to public safety on the I-395 express lanes in Arlington. Pictured is a touch screen that law enforcement uses to communicate with the vehicle. (Photo courtesy of Transurban North America)
Researchers at Virginia Tech’s Transportation Institute will soon hand the keys to lawmakers and developers, after creating a blueprint for how self-driving vehicles can safely interact with other cars on state roadways.
In 2019, Virginia Tech was awarded a $7.5 million grant from the U.S. Federal Highway Administration to gather data on improving autonomous driving vehicles (ADS) in roadway scenarios – including navigating safely around first responders, construction workers and work zones – amid increases in self-driving vehicle crashes.
“Our goal is just to make transportation better,” said Zac Doerzaph, executive director of VTTI, after an Oct. 11 presentation in Northern Virginia on the closed Interstate 395 express lanes.
The project was completed in partnership with the Virginia Department of Transportation, Crash Avoidance Metrics Partners LLC, a consortium of vehicle manufacturers and global toll operator Transurban North America, which oversees the Interstates 495, 95, and 395 express lanes in the Washington, D.C. metropolitan region.
Issues with ADS
While autonomous driving vehicles, or self-driving vehicles, rely on artificial intelligence and can operate without a driver, the vehicles and their developers have faced criticism for their increase in crashes nationwide. Data from the National Highway Traffic Safety Administration, an entity under the U.S. Department of Transportation, shows that Virginia recorded 13 crashes with SAE Level 2 Advanced Driver Assistance Systems, which increased to 31 through August. SAE International, which is a global association of engineers and technical experts, developed the levels based on a self-driving vehicle’s capabilities.
In June 2021, the National Highway Traffic Safety Administration, an entity under the U.S. Department of Transportation, began requiring manufacturers and operators to report crashes involving their self-driving vehicles.
“The U.S. Department of Transportation’s top priority is safety,” the highway administration said in a 2019 statement. “Automation offers the potential to improve safety for vehicle operators, occupants, and other travelers sharing the road.”
Not every automated driving system is the same and varies based on their offerings.
According to the safety administration, the basic ADS can perform without driver involvement, but is still being tested on public roads and is not available for consumer purchase. A Level 2 advanced driver assistance system provides both speed and steering input, but requires the human driver to remain engaged in driving at all times.
Another issue with self-driving vehicles is their lack of response to commands because they rely on wireless connectivity.
On Oct. 24, autonomous car company Cruise was ordered to remove all of its vehicles from roads in San Francisco, after the California Department of Motor Vehicles and the state’s public utilities commission determined the vehicles caused “unreasonable risk” to public safety.
The suspension was handed down after the General Motors-owned tech startup failed to disclose the full details of an Oct. 2 collision that trapped a pedestrian underneath an autonomous vehicle.
For four years, the project team worked to create a blueprint for autonomous vehicle developers. The group consisted of equipment manufacturers and public safety and services partners to help address the challenges of safely operating autonomous vehicles in real-world scenarios, according to Tech’s Transportation Institute.
Researchers said one of the challenges for self-driving vehicles is the need for more real-time and distance information, which the vehicles need to optimally operate.
The group then developed a software platform that could provide information such as the traffic conditions, work zones, and potential incidents to the vehicle.
Transurban supported the project by allowing researchers to use the I-395 express lanes in Northern Virginia to identify solutions for self-driving vehicles in several scenarios. The lanes are also equipped with roadside sensors to help the institute’s Level 4 Ford F-150 autonomous vehicle, equipped with sensors and cameras.
“The future of how we get from point A to point B is changing every day. Transurban and our partners know we must continually innovate to keep travelers safer as technology enables the next generation of transportation,” said Mike Discenza, acting president of Transurban North America. “Transurban’s proprietary, embedded technologies and closed environment on the 395 Express Lanes enable the perfect testing ground for infrastructure that talks to vehicles and vice versa — making roads safer as travel turns autonomous.”
The Crash Avoidance Metrics Partners LLC helped develop a list of 18 situations for the research. Some included detecting upcoming hazards and navigating through construction workers and work zones.
“A lot of the scenarios that we developed are impactful,” said Joe McLaine, general safety manager for General Motors and chair of the CAMP LLC. “We believe that solving them and showing technological innovation and movement towards implementing those is a great step and the reason why we’re a part of this project.”
One situation of note was when the F-150 responded to law enforcement by pulling over to the far right lane on the closed express lane. When the Virginia State Trooper car pulled closer to the ADS and flashed its lights, the vehicle turned its right blinker on and when appropriate, moved from the middle lane to the emergency lane.
The trooper exited his vehicle, approached the ADS on the right passenger side and used a touch screen on the window to communicate with the vehicle’s fleet manager or a human to resolve any issue. After gathering the necessary information, the trooper proceeded back to his vehicle. The ADS then turned on its left blinker and safely merged back into its previous lane.
Researchers said the early stages of the project required a lot of trust because the autonomous vehicle can come to a more abrupt stop compared to a human driver.
Last year, the Virginia Department of Transportation told the Mercury that Virginia has no law written for autonomous vehicles. The agency published a strategic plan for operating autonomous vehicles in 2020. VDOT expects deployment of the vehicles to impact safety and mobility for drivers, and create opportunities for economic development.
Virginia has already started making space for self-driving vehicles. According to SmartCities Dive, Interstate 66 in Northern Virginia reopened with lanes to support self-driving cars last year.
Doerzaph said the study is expected to help lawmakers with future policy decisions, one of which could be in establishing an interoperable connected ecosystem for transportation.
Doerzaph said the Manual on Uniform Traffic Control Devices, which the Federal Highway Administration created, provided the best practices for the way roadways are designed. The manual also helped during the study by identifying what work zones are easier to address for ADS than others.
Mike Mollenhauer, division director of technology implementation at VTTI, said the sensors provided by Transurban helped the self-driving vehicle understand the road conditions, but he is unsure if vehicles will need them in the future. Sensors “can really make the experience better,” he said.
“I think, at some level, you have to have some ability to have infrastructure play a role in this deployment, but I don’t know that you have to have [it] as competent of a system as what Transurban operates here,” Mollenhauer said.
“Certainly for early deployment, it’s a benefit,” he added, “but because you want that additional information, you can potentially do it safer than you could otherwise.”
Virginia Tech will publish its findings online in early 2024, including videos and photos, according to Mollenhauer.
GET THE MORNING HEADLINES DELIVERED TO YOUR INBOX
Our stories may be republished online or in print under Creative Commons license CC BY-NC-ND 4.0. We ask that you edit only for style or to shorten, provide proper attribution and link to our web site. Please see our republishing guidelines for use of photos and graphics.