Engineering Duo's AI Project Boosts Highway Safety

Vanderbilt University

When Jonathan Sprinkle aced his driver's education course as a teenager in his small East Tennessee town, his teacher jokingly called him an "old lady driver," and the oldest of five siblings graciously took it as a compliment.

He combined that focus and care for others with sharp intelligence, which opened doors for him in school to be an early adopter of what was then rare-desktop computers.

Sprinkle had a transformative experience when a computer-based milling machine exhibit came to his high school. It was intended to show vocation technology students advanced manufacturing and milling techniques with computer-aided drafting.

"In 1994 this was pretty revolutionary, because it meant that instead of hand-controlled milling, the computer could gain degrees of consistency that trained humans could not," he said.

Sprinkle and a friend found that by hacking that milling machine they could create 3-D objects.

"This simple act of integrating an existing working system with some high-level logic laid the foundations of my interest in computing to interact with physical things," he said.

Fast forward to today and Jonathan Sprinkle is professor of computer science, civil and environmental engineering, and of electrical and computer engineering at Vanderbilt.

COUNTING CARS

Dan Work had a near-miss experience as an exhilarated new teen driver, cruising out of his local DMV near Columbus, Ohio, with his parents and sister in tow.

"That wasn't my best left turn," Work said, laughing. But his energy and enthusiasm soon became the fuel of a visionary civil engineer's imagination.

Work interned in his late teens with his local city government. The engineers had him either scanning drafts into a computer or sitting in a lawn chair near an intersection counting cars to see how the traffic light timing should be altered.

"I kept thinking, there's got to be a better way to ease traffic," Work said.

As the world advanced from paper maps to GPS systems, Work immersed himself in the rich civil engineering research tied to traffic.

"In civil engineering, no one notices the science if everything works," Work said. "It's fascinating and super complex, and I like that I can quickly see the practical results of what we're doing, especially with traffic research."

Today, Dan Work is a professor of civil and environmental engineering, and of computer science, and part of the Institute for Software Integrated Systems at Vanderbilt, which works in cyber-physical systems, security, education, model-integrated computing and distributed large-scale system integration.

These two Ph.D. powerhouses have come together to combine their contrasting personalities and complementary expertise in engineering, computer programming and artificial intelligence to tackle the vital issues of driving safety, fuel economy and traffic jams.

SAFETY IN THE SKY

While Work was on the ground studying traffic, Sprinkle focused his expertise on safety in the sky-experimenting with AI software that could safely control pilotless aircraft.

"I started asking myself, how can I do research that will positively influence 10,000 people or more? I wanted every project to have that potential so that many people could benefit from it," said Sprinkle, who is also chair of the computer science department and a member of the executive council of the Institute for Software Integrated Systems at Vanderbilt.

Spurred by a grand challenge project launched by the U.S. Government's Defense Advanced Research Projects Agency, or DARPA, Sprinkle took his autonomous aircraft work and adapted it to vehicles.

JOINING FORCES

Work and two mathematician colleagues had a vision that using autonomous vehicles in some way could help fuel efficiency and instabilities in traffic flow. But they didn't have expertise in autonomous vehicles.

Sprinkle did. And that's how this dynamic duo came together.

They had met years earlier at the University of California, Berkeley, so Work reached out to Sprinkle. "I called Jonathan and said, 'you're probably the only person I know that would even consider working with this group of misfit toys. Would you like to teach us everything we don't know about autonomous vehicles? And Jonathan, being a super brilliant and kind human being, not only intellectually started talking through how to make this idea come to life in a deep and meaningful way, but he said he thought it would be fun. And the rest is history."

FROM DUCT TAPE TO DISCOVERY

The essence of the first research project was to see if one semiautonomous car, with careful control of its speed, could positively impact the speeding up and braking of other cars by dampening traffic waves.

The study started with lots of technical hiccups, duct tape and even pool noodles to fix a temperamental light sensor.

"You see the final research video and everything's working wonderfully, and it's just like, 'oh, these people have this great idea,'" Work said. "But the process felt more like 'nothing works until everything works.'"

Through all the learning curves, the 20-car study was a huge success, showing that autonomous software that set just one car at a consistent speed could smooth traffic waves for all the cars.

GO BIG OR GO HOME

While Work, Sprinkle and teammates celebrated their discovery, fellow researchers and car industry experts wanted more.

"We knew we were going to have to go much bigger, be much bolder," Sprinkle said.

"We were going to have to ramp up the experiment on an open multilane freeway. And it's going to take a lot more resources. And that's when the freakout moment starts," Work said.

100-CAR CHALLENGE

In November 2022, their research project dubbed the CIRCLES 100-car challenge would involve testing an AI-powered cruise control system on 100 Nissan Rogue vehicles interspersed in traffic on an active highway-over multiple days.

"This is now not only a research project, but it's also an organizational project, a logistics project, a safety project, an insurance project, a procurement project, a networking project, an IT project," Work said.

Work and Sprinkle needed help-lots of help.

ESSENTIAL PARTNERS

They built up a strong team of researchers in all levels of engineering and computer programming.

"I started working with Dr. Sprinkle at the University of Arizona and stayed with him through my Ph.D. in electrical and computer engineering, then came with him to Vanderbilt," said Matthew Bunting, research scientist at the Institute for Software Integrated Systems.

Among other things, Bunting helped build and install the mechanisms that adapted the cars into semiautonomous vehicles.

"I've seen what Dr. Sprinkle has done. I've seen how he treats people. I know how he treats me. He is an incredible person that I wanted to stay with," Bunting said.

They also partnered with researchers at UC Berkeley, Temple University and Rutgers University, establishing the CIRCLES Consortium, which received generous funding from the National Science Foundation's Cyber-Physical Systems grant.

Nissan North America donated the cars used in the study and allowed Sprinkle and his team of software engineers to install their own cruise controllers.

"The partnerships were essential, with safety as our number one priority," Sprinkle said.

One of their most essential partners was the Tennessee Department of Transportation-who believed in Work and Sprinkle's research and vision so much that the state built a four-mile-long testbed for traffic management and automated vehicle technologies.

Jonathan Sprinkle talks to a volunteer before the experiment. (Vanderbilt University)
Jonathan Sprinkle talks to a volunteer before the experiment. (Vanderbilt University)
I-24 MOTION cameras built in the Vanderbilt engineering lab (Vanderbilt University)
I-24 MOTION cameras built in the Vanderbilt engineering lab (Vanderbilt University)
Volunteer drivers get a review of the research project. (Vanderbilt University)
Volunteer drivers get a review of the research project. (Vanderbilt University)

FIRST-EVER SMART HIGHWAY

The I-24 MOTION "smart highway" is four miles long and equipped with 300 4K digital sensors capable of logging 260 million vehicle-miles of data per year. It is the only real-world automotive testing environment of its kind in the world.

"We've been just extremely fortunate that TDOT has walked this walk with us and given us opportunities that other researchers would dream to have in terms of the time that they've spent with us and the resources they've provided," Work said.

Dan Work shows the mechanism connecting the AI-powered cruise control system for the research project. (Vanderbilt University)
Dan Work shows the mechanism connecting the AI-powered cruise control system for the research project. (Vanderbilt University)
The CIRCLES 100-car challenge involved testing an AI-powered cruise control system on 100 donated Nissan Rogue vehicles interspersed in traffic on an active highway. (Vanderbilt University)
The CIRCLES 100-car challenge involved testing an AI-powered cruise control system on 100 donated Nissan Rogue vehicles interspersed in traffic on an active highway. (Vanderbilt University)

VANDERBILT SUPPORT

Vanderbilt played a pivotal role in the project, acting as the problem-solving hub. From facilitating connections between researchers and key partners to ensuring seamless Wi-Fi connectivity beyond campus, dozens of logistical hurdles were tackled.

"So many times at Vanderbilt a conversation would start with the phrase, 'All we need to do is,'" Sprinkle said. "It was this mentality of 'it doesn't matter that it seems impossible-we're going to sit down and figure out how we can do it.'"

INSPIRED COLLABORATION

After years of planning, the 100-car challenge project transcended from a study to an inspired collaboration. Work and Sprinkle had dozens of friends, colleagues and family members as safety drivers and volunteers in the experiment.

"Everyone in the study had to wear a specific safety vest with a number on it. My son, who was 14 at the time, sat at a sewing machine for three hours and sewed 150 patches on these vests," Sprinkle said proudly. "It was kind of unreal."

Work echoed this sentiment with a heartwarming anecdote about a highly respected colleague, now vice chancellor for research at Rutgers University.

"Benedetto Piccoli, one of the most famous mathematicians in traffic flow theory on the planet, went to Costco with my mom and dad to get supplies so we could clean all the borrowed vehicles," Work said. "That was such a full circle moment and illustrated how special the project was beyond the scientific impact that it had."

PARTNERS IN SAFETY

With the massive success in predictive traffic management from the I-24 MOTION highway system and solid automaker partnerships, this duo is ramping up new AI-focused driving safety research projects.

What will stay consistent is the admiration they have for each other-so much so that they chose to have offices only separated by a coffee maker and an idea-covered white board, with a few puzzles and jokes written along the edges.

"Dan is an incredible friend who has been there to support me when I push myself too hard, and who reminds me about what is really important," Sprinkle said. "Because of the close-knit relationship and camaraderie that we have, I know that I've got someone every day who I can bounce ideas off of. Someone to make new discoveries with that could transform society."

"Jonathan is just a special human being in a million different ways," Work said. "He gives me inspiration for how to be not only an excellent researcher, but an excellent husband, an excellent father, an excellent community member. And I think that's not just true for me, but for basically everybody that gets a chance to work with him."

By Amy Wolf

/Public Release. This material from the originating organization/author(s) might be of the point-in-time nature, and edited for clarity, style and length. Mirage.News does not take institutional positions or sides, and all views, positions, and conclusions expressed herein are solely those of the author(s).View in full here.