U of A Self-Driving Car

Eight University of Arizona engineering and computer science students from across the nation have been working this summer to advance driverless car technology. Using the University of Arizona’s Cognitive and Autonomous Test – or CAT – vehicle, they have been conducting research on autonomous vehicles. One of those students, Carlos Asuncion, will talk about his work, along with Jonathan Sprinkle, a professor from the U of A Department of Electrical and Computer Engineering.

Transcript:

Ted Simons: Good evening, and welcome to "Arizona Horizon." I'm Ted Simons.

Ted Simons: The Maricopa County sheriff's office is receiving an increased number of complaints this year, but officials say that's to be expected with the department asking for increased community input as part after federal court order to stop racial profiling. The first half of the year the sheriff's office has received complaints compared to for all of last year.

Ted Simons: And a candidate running for Congress in Congressional District One announced he has cancer, Republican Adam Kwasman says he was diagnosed with a year ago with chronic lymphocytic leukemia, a slow-growing blood cancer. Kwasman says he'll remain in the race, he says the cancer was caught early, he has no symptoms and he likely will not require treatment for at least years.

Ted Simons: Tonight's edition of Arizona Technology and Innovation looks at how students have been working this summer to advance driverless car technology. The research is being conducted at the University of Arizona and includes use of autonomous test vehicle provided by the U of A. Here to talk about the research is Jonathan Sprinkle, a professor from the U of A department of electrical and computer engineering, and Carlos Asuncion, a student who worked on that technology. Good to have you both here. Thanks for joining us. OK. Driverless car technology. What exactly are we talking about here?

Jonathan Sprinkle: We're talking about technology that takes the driver out of the loop for some tasks. Quite a few of the injuries and deaths that happen every year in the U.S. happen in vehicles, and they happen at speeds that are not just fast speeds. Even under 25 miles an hour there's a lot of dangerous things that happen.

Ted Simons: And we had students from across the country, what were they working on? What were they advancing?

Jonathan Sprinkle: They were thinking about small projects they could do that would improve the state of the art. Some of the teams were focused on how high school students would be able to use the autonomous car in the high school, some of the teams were interested in decreasing the cost of the sensors that belong on autonomous vehicles. That's an important consideration for consumer devices.

Ted Simons: Explain why that's important, these sensors.

Jonathan Sprinkle: Some of them can run as high as $50-80,000. And that's a significant chunk of money to pay when your vehicle may not be worth that much. Not to mention I want to have maybe more than one sensor in case it fails. So we'd like to be able to have redundancy.

Ted Simons: It's not like we read about this Google car people sit around and drive all over, the Valley up there, this is a little different and a little less costly?

Jonathan Sprinkle: Well, we still have the expensive sensors to test. What we'd like to be able to do is say we can do the same thing for much less money that this expensive sensor can do. We need to compare it to the expensive sensor in order to do that.

Ted Simons: What specifically did you work on, Carlos?

Carlos Asuncion: I worked on seeing if we need the high-end sensor for the car to safely operate.

Ted Simons: What did you find out?

Carlos Asuncion: We found the sensor we were using has 64 lasers that it uses to scan the area to detect obstacles and we determined that for different driving situations, you might not need the lasers in order to operate safely.

Ted Simons: So maybe you need 32, or zero, or --

Carlos Asuncion: We found that for like highway setting, 16 , and for maybe parking lot settings, where there's pedestrians and smaller objects, you need to avoid, as small as 32.

Ted Simons: Describe the process. Like a typical day of research for you this summer. What did you do?

Carlos Asuncion: Well, we took the data we get from the high-end sensor and we would manipulate it to simulate other sensors with, say, lower accuracy, lower resolution. And we compressed that to a two-dimensional height map that would allow us to compare the different sensors.

Ted Simons: Is this -- Are these sensors looking for other cars on the road, telephone poles, curves in the highway, or is there a mapping technology that allows these sensors to understand what's happening out there in the first place?

Carlos Asuncion: The way the sensor operates, it catches -- It takes a sweep and gives you an image, a 3D image of the surroundings, and then from there you need to run software in order to detect, OK, this object is a car, this is a pole, avoid those areas.

Ted Simons: And how far along are we with this particular technology?

Jonathan Sprinkle: I think it's getting closer and closer. So companies like Google and other companies that take many of these devices or produce these devices are very sophisticated in how they can consume this data and infer information about it. What we're interested in is in being able to take something and just say if we have this, we could drive in a parking lot, or your car could get gas on its own while you were waiting at work. Little things like that don't seem like a lot, but I think that's the near term future of autonomy.

Ted Simons: That seems like quite a bit, actually. And a full-size car now, this test vehicle was a full-size vehicle?

Jonathan Sprinkle: That's right. And that's something that makes it both exciting and also extremely frightening. It's one thing to have a remote controlled car and be able to say you can make it do cool stuff, but it's dangerous to do it the wrong way.

Ted Simons: I would imagine, as we're watching this video of this car moving around, the steering wheel is moving, you can't do much testing on a roadway because everyone would look at the car and you'd have accidents because people are looking at the driver with no driver.

Jonathan Sprinkle: Yeah.

Ted Simons: So the parking lot is basically as far as you've gone so far.

Jonathan Sprinkle: To some degree. This lot, we -- Our gracious sponsors at the University of Arizona allow us to drive in this lot any time we want. We have near road conditions to use if we can close those off further south of town. But we've been focusing on showing proof of concept, and we would in the academic world publish a paper and let industry take it from there.

Ted Simons: Carlos was talking about using the sensor technology, did different students concentrate on different things?

Jonathan Sprinkle: Each student group was focused on a different piece of the puzzle. And that's to give the students who were part of the program the experience of being in graduate school and thinking about big problems and society that they might be able to work on.

Ted Simons: Give us some examples of what they were working on.

Jonathan Sprinkle: One of the great examples was something called switched control. When you're driving on the highway, you drive differently than when you're driving in a parking lot. And when you're driving in a parking lot with no cars you drive differently than a parking lot with lots of active traffic and people walking their dog. So we need software that can switch back and forth to different designs and different controllers depending on things that detects in the environment.

Ted Simons: Interesting. Is that advance something getting there?

Jonathan Sprinkle: Absolutely. Part of it is not just being able to do it, but to think on a high enough level and to essentially be able to explain your design in a picture is really important. And that's what they were able to do.

Ted Simons: What got you involved in this? Where did this start with you?

Carlos Asuncion I got really interested in control systems and I was doing a different research program, and I wasn't really into it anymore. So then I did a quick search on the NSF website and found this program.

Ted Simons: National science foundation. Basically a summer program, correct?

Carlos Asuncion: M-hmm.

Ted Simons: And again, we're talking about robotics, autonomous vehicles. But this technology can be used in other things. Correct?

Jonathan Sprinkle: Yes. Absolutely. It's not exactly clear what we think of or what we invent today how that's going to be used. That's why it's important for people doing research in this area to publish what they're working on. That to some degree is what separates us from the work going on by Google and other people. Of the work they do which is absolutely incredible, I don't want to say anything bad about the work they're doing, but it's difficult for people outside of their area to get access to it.

Ted Simons: When we're talking about medical monitoring, we're talking about border enforcement, these things, that's -- That all play as part.

Jonathan Sprinkle: Yeah, and in fact we've been able to apply different kinds of results to people who drive assisted wheelchairs or electric wheelchairs to prevent rollover and driving over the curb and other things that cause serious injury.

Ted Simons: As far as you're concerned, you're at Berkeley, correct?

Carlos Asuncion: Correct.

Ted Simons: Where do you want to take this? What's next for you?

Carlos Asuncion: Probably looking into a different research program at my home school and continue research in autonomous systems.

Ted Simons: So that's pretty exciting for you.

Carlos Asuncion: Yeah.

Ted Simons: Was that always the way? I know you looked up this particular program, you had a good robotics program at your high school, so that was instilled in you, huh?

Carlos Asuncion: M-hmm.

Ted Simons: As far as what's next in this particular technology, you're the expert. You've won all the awards, what is next in autonomous technology?

Jonathan Sprinkle: I think what we'll see is more and more autonomy creeping into today's vehicles, and we're even seeing it now on commercials on television about vehicles that have lane departure warnings and self-parking. That's going to become more and more common. Where it will be interesting is when we start to apply that as you said around other vehicles with people looking out at the spectacle that's going on. It will be interesting to see where things go from there and that's almost a humanities question.

Ted Simons: Indeed. Because there is a societal impact. You don't work in a vacuum, this has got to work among the people, the other drivers.

Jonathan Sprinkle: Exactly. And I think that's something that's often forgotten about engineering. We have to think about building things that improve society. So you do need some societal context to do that.

Ted Simons: Congratulations on your honors and your wards, and for teaching these young people, and congratulations to you, and good luck.

Carlos Asuncion: Thank you.

Jonathan Sprinkle: Thank you.