When cars talk
Photo by Owen Fitter April 09, 2021
In the future, self-driving cars will communicate to save fuel, improve safety
There’s a beautiful physics to the synchronized movement of a flock of birds, or school of fish, travelling with mesmerizing precision to take advantage of aerodynamic factors that save energy.
What if cars could do the same?
That’s the question being asked by Andreas Malikopoulos, the Terri Connor Kelly and John Kelly Career Development Associate Professor in the Department of Mechanical Engineering and the Director of the Sociotechnical Systems Center at the University of Delaware (UD). Before joining UD in 2017, he was the Deputy Director and the Lead of the Sustainable Mobility Theme of the Urban Dynamics Institute at Oak Ridge National Laboratory, and a Senior Researcher with General Motors Global Research and Development. At UD, he is developing control and learning algorithms that will fundamentally change how cars interact on the road in order to maximize fuel efficiency.
The answer, of course, is to minimize the control of humans and let computers take the wheel.
Malikopoulos researches what are known as connected and automated vehicles, basically autonomous cars that use sensors, cameras and advanced control algorithms to adjust their operation to changing conditions with minimal or no driver input while also sharing with other vehicles information about location, speed, direction and so forth.
Malikopoulos calls it the “Internet of Cars.”
In partnership with the University of Michigan, Boston University, Robert Bosch GmbH, and Oak Ridge National Laboratory, Malikopoulos is the principal investigator of a $4.2 million, three-year project funded by the Advanced Research Projects Agency for Energy (ARPA-E) through its NEXT-Generation Energy Technologies for Connected and Automated On-Road Vehicles (NEXTCAR) program.
After extensive testing on simulators at UD and on Mcity, an experimental test track at the University of Michigan, using an Audi A3 e-tron, Malikopoulos’ team exceeded the program goal and achieved energy savings of 25% by the end of the grant period in December 2020. (Video of the work on the test track can be seen here.)
In an interview with UDaily, Malikopoulos explained how connected and automated vehicles will change driving forever by putting humans in the passenger seat.
Q: What are connected automated vehicles?
Malikopoulos: When we say connected and automated vehicles, it means vehicles with some degree of automation (can be fully autonomous) that can exchange information — like location, speed and acceleration — and make decisions that will improve safety and efficiency.
Q: What does this look like in practice?
Malikopoulos: For example, if I am approaching an intersection in a connected automated vehicle, and if there are other vehicles approaching the same intersection, then my vehicle can accelerate or decelerate and cross the intersection without coming to a conflict with other vehicles. If all the vehicles have the same information, then they can coordinate with each other to avoid stop-and-go driving. And if we can avoid stop-and-go driving, then we can improve energy efficiency while reducing travel time.
Q: What does this look and feel like for the driver in the connected automated vehicle? Are we talking about more gradual acceleration and deceleration?
Malikopoulos: Absolutely. Imagine that you are driving a connected automated vehicle and it starts to gradually decelerate. You might wonder why this is happening. Not yet visible to you is an intersection with stopped vehicles. The connected automated vehicle gets information from the other vehicles at the intersection and starts decelerating so by the time you approach the intersection there are no other vehicles, and you can cross the intersection without coming to a full stop.
Q: This isn’t just about self-driving cars getting from Point A to Point B; it’s about how many vehicles can get from their Point As to Point Bs while using as little fuel as possible?
Malikopoulos: That’s a great point. Even if you have a fully autonomous vehicle, if this vehicle is not connected to other vehicles — if it doesn't send or receive information — then you don't really have benefits. You just have a computerized autonomous vehicle that can make decisions, but eventually it will have to stop at an intersection just like human-driven vehicles. The difference is the connectivity.
Q: How did you test this? Were there any safety concerns?
Malikopoulos: We went to the University of Michigan to run experiments because they have a test-track facility outfitted with augmented reality features. So you can put a physical entity onto the track, the Audi in this case, but then there are virtual vehicles that you cannot see but exist in the simulation environment. These virtual vehicles are communicating with the Audi. So although you see the Audi alone on the track, essentially the Audi is among other virtual vehicles that broadcast information and coordinates with the other vehicles. For example, when the Audi comes to a stop, it's because there is another virtual vehicle ahead in the simulated environment. With this framework, we were able to do experiments safely. If something goes wrong, it's just virtual vehicles. There's no crash.
Q: What did you learn from these experiments? What were the results?
Malikopoulos: We eventually exceeded the goal designated by our sponsor, ARPA-E. The goal was to improve the fuel economy of an Audi A3 etron by 20%, and our team reached 25% improvement. Think of it this way. If every vehicle in the transportation network were 25% more efficient, then we could drastically reduce oil consumption in the U.S.
Q: Any chance consumers will see this technology on the road soon?
Malikopoulos: Our partner, Bosch, has integrated the technologies that we developed into their product portfolio and, hopefully, they will become even more widely available in the next generation of vehicles. But it’s complicated. Even when the technology has matured to the extent that we now have with connected and automated vehicles, there are still several additional steps, like creating the infrastructure to make this possible on a larger scale. Imagine small communication networks, like an internet of cars. Each vehicle will have its own network connected to other vehicles. So some of these technologies will not be widely available until we have the appropriate infrastructure.
Q: What are the next steps? What are some additional problems that need to be resolved?
Malikopoulos: We definitely have many challenges left, one of which is to make sure that we can have a safe coexistence between human-driven vehicles and connected automated vehicles. A connected automated vehicle is operated by a computer, so you can't really have an unexpected event. But humans can do anything, like accelerate when traffic is slowing down, which is what makes it so difficult to predict driving behavior. So the next step is bridging this gap.