Are We There Yet?
When Ümit Özgüner explains his research on autonomous driving, he says he has a mental image of Star Trek’s USS Enterprise navigating around wormholes, Romulan Warbirds and asteroids.
In Özgüner’s vision, however, the starship wouldn’t be under the command of Captain Kirk — or any captain, for that matter — to boldly go where no man has gone before.
Recognized internationally for his expertise in autonomous vehicles research, Özgüner, professor of electrical and computer engineering, and four College of Engineering colleagues are working under a $1.49 million National Science Foundation grant to examine challenges of autonomous driving in mixed-traffic urban environments.
Anticipating a world where software and virtual environments are intermingled with physical systems as we know them, they are looking for solutions to problems that could arise if autonomous vehicles shared the roadways with human drivers.
Specifically, their work addresses three areas:
- Developing ways in which relatively simple solutions, such as programming autonomous vehicles to negotiate one intersection, can be expanded and applied, say, to the traffic grid in an entire town
- Programming an autonomous car or robot to have robust behavior so it can deal with unforeseen behaviors of human drivers
- Allocating and distributing computational power for autonomous vehicles in the process of understanding the surroundings.
“It may not be that hard to sense exactly what you have around you,” explains Özgüner, who is director of the Control and Intelligent Transportation Research Lab at Ohio State. “It is much harder to understand what those things around you may be doing.”
For instance, if a car stops in front of an autonomous vehicle, how would the autonomous vehicle decide whether to go around it, such as would be feasible on an open roadway, or to wait, because perhaps the front vehicle is at a red traffic light?
“We humans solve this in context without even knowing it,” Özgüner says. “An autonomous vehicle might be able to sense a ball rolling into a roadway. But could it then figure out that there may be a child running out into the street after it?”
Özgüner’s NSF team includes Professor Füsun Özgüner; Associate Professor Ashok Krishnamurthy, interim director of the Ohio Supercomputer Center; and Keith Redmill, a research scientist, all in electrical and computer engineering; and computer science and engineering faculty Bruce Weide and Paolo Sivilotti, as well as graduate students from both departments.
On the application front, Özgüner and his colleagues are working with Honda on determining driver behavior at intersections in the hopes of developing warning devices. In another project, with Ford, they are investigating use of autonomous technologies in plug-in hybrid electric vehicles. And in collaboration with a researcher from the Japanese government, they are investigating fault tolerance of automotive systems.
“The intent of Japan is to apply this to automated truck convoys that will save energy,” Özgüner says, explaining that the first truck in the convoy would have a human driver, but trucks behind it would be autonomously operated. The vehicles in the convoy would travel very quickly and closely together to take advantage of drafting.
“It’s not that human drivers can’t do that, but if you’re driving 65 miles per hour, 5 feet behind another truck, you can’t do that for too long, so you’d better have it automated,” he says.
Özgüner is participating with a team of researchers in Turkey in MAGIC 2010, the Multi Autonomous Ground-robotic International Challenge hosted by the Australia and U.S. Departments of Defense to develop next-generation fully autonomous ground vehicle systems that can be deployed effectively in military operations and civilian emergency situations. The team, Cappadocia, in July was among six selected as finalists for the November contest in Australia.
In addition to investigating specific challenges of autonomous driving, Özgüner and his colleagues are keeping watch on sensing and intelligent computing technologies and trends that will affect their research many years down the road.
“Certainly, in the last 10 years, GPS has become more prevalent and popular, and wireless communication, pushed by cell phones, has become more common,” he says, “so we expect cars in the future to be communicating with each other, definitely.”
Ümit Özgüner, (614) 292-5940, firstname.lastname@example.org