Chris Gerdes is passionate about what he does, and what he does will knock your socks off. As the Director of the Center of Automotive Research at Stanford (CARS), he spends a good chunk of his time around race cars… race cars with no one at the wheel. Yes, you heard us right. Gerdes’ research centers on autonomous vehicles — unmanned machines that drive themselves. The goal? To create safer, forward-thinking automobiles that will change the way we think about transportation. Take a look at some of the videos; we think he’s well on his way to meeting that goal.




Oh, and if you’re an innovator and think you’ve got an idea — like Chris’s — that will change the world, check out; they’re providing two outstanding innovators with $50,000 to jumpstart their ideas. (Plus a new Altima. Plus a Kickstarter campaign.) It could be you, so don’t hesitate! You might be our next BR!NKer.


First of all, can you explain to me how autonomous cars — like Google’s self-driving car and the cars you’re currently creating — work? I’m ashamed to say that I don’t know!


Self-driving cars basically need three things — sensors for understanding their surroundings, algorithms for deciding what actions to take based on the sensor measurements, and actuators that can steer, brake, and drive the car to perform those actions. The number and type of sensors used vary according to the driving task. With Shelley, our autonomous Audi TTS that we use for racing, a lot of our focus is on the car’s motion itself. Our primary sensor suite is a combination of a two-antenna differential GPS system, accelerometers, and gyroscopes that tell us exactly what the car is doing. We can identify its location on the track to within a couple of centimeters and measure how much the tires are slipping relative to the road, which tells us how much of the tire’s capabilities we are using. Cars designed to operate in traffic use additional systems like radar and scanning lasers to map out the location of obstacles in the world.


With all of this information available, the next step is to decide what to do with it. That’s really the heart of the research in this field — developing rules or algorithms that can process all of the sensor data and translate it into a desired path and actuator commands for the car. There are a lot of different approaches to this and our lab focuses on using as much information about the underlying physics as possible. This enables us to quickly port solutions to different cars and handle changing conditions easily.


Finally, there is the need for actuators that can control the vehicle. With modern cars, this is by far the easiest part, since most come from the factory with the ability to do these things. Virtually all new cars have electronic throttle and brake control and can accelerate and decelerate in response to computer commands; cars with electric power steering can also steer themselves based on a simple message sent to the steering control unit.



Shelley, the self-driving Audi TTS


I would describe what you’re doing, in layman’s terms, as “building driverless race cars.” What is the goal of your current research with the Center for Automotive Research (CARS) at Stanford? (Besides creating something really, really cool.) It looks like there are two facets: instrumenting the cars, and instrumenting race car drivers themselves.


At the end of the day, our goal is to develop innovations that can lead to safer cars on the road. When you look at conditions where accidents — and particularly fatal accidents — occur, a lot of times you find the car stretched to the limits of what it can do. The car spins due to a lack of grip on any icy road or slides off the road because it was going too fast to make a turn. These are the same sort of situations that race car drivers handle every second they are on the track.


Working with professional race car drivers, you quickly realize how good humans can be at taking the car to its absolute physical limits. There simply isn’t that much room for improvement in the lap times they produce. We want to understand how they are doing this by measuring their actions on the track and even their brain waves as they are driving. This tells us which of their skills come from making good decisions and which are reflexes built up over years.


We are also trying to encode some of their tremendous abilities in robotic cars by developing algorithms that can achieve the same performance as humans on the track. The objective isn’t to create robotic racing, but to study racing as a way of getting insight into obstacle avoidance and other challenges the car sees in everyday driving. Since most of us in the group love to drive, we’re not sure whether the best solution is for the driver to simply be a passenger in an autonomous car or whether a better balance exists. Given the wonderful power of the human brain and how well an engaged human driver performs, it seems a great loss to disconnect people entirely. Finding a way for the driver and machine to work cooperatively in accident situations is a major research challenge for us.


When BR!NK first contacted you for this interview, you mentioned you had to be up early the next morning to be on the track. What is an average workday like for you? I’m imagining engineers sitting with computers on the side of the track, with an occasional explosion occurring in the background.


Well, we certainly go to great lengths to minimize the number of explosions involved with our testing, although revving engines and screeching tires are very common. Your image of engineers sitting with computers by the side of the track is spot on, though. There’s a lot of time looking at data, punctuated by moments of intense drama during the actual testing.


When we are instrumenting human drivers during a race, there is a flurry of activity in the paddock right before the car heads to the pits. We have just a few minutes usually to check out all of the sensors, wire up the driver, and get the cameras going before the race begins. After that, we get to be spectators for a while and cheer on our drivers (while occasionally counting the number of antennas on the roof to make sure it hasn’t changed).


When we run Shelley, our autonomous car, we go through a countdown procedure to make sure all systems are working and our spotters around the track are in radio contact and able to stop the car in the event of an emergency. We then send the car out to lap, collectively holding our breath that everything goes as planned. When the car returns to the pits, there is a moment of celebration (particularly when our lap time drops), and then it is back to the computers to come up with the next thing to try.


I spend far more days on campus than I do on the track, so track days aren’t really typical. On campus, I really enjoy teaching, particularly when students get those “aha!” moments understanding how things work. Lectures are fun, but laboratory experiences such as dissecting a transmission or racing around the go kart track are even better. Other days, the administrative tasks really pile up, and I dream of ducking out the window of my office like Indiana Jones in the Last Crusade and heading off to the track.



Shelley in action. No explosions, but plenty of screeching tires.


You mention in your TEDx talk that the autonomous race car can operate equally well in all weather conditions — sunshine, wind, and rain. As I driver, I know I have to change my driving depending on external factors — how can an autonomous car do that?


Adapting to changing conditions is extremely important. The car is connected to the road through the friction in four little patches about the size of your hand. When these patches get wet or icy, the available friction there drops, placing severe limits on how fast the car can corner or brake. Like the best race car drivers, our autonomous car must keep looking several turns ahead, calculating how fast it can take each turn and then figuring out the right point to brake. The correct speed and brake point depend on the friction available, so the car needs to be able to estimate this. Race car drivers get a lot of this information by feeling the road through the steering wheel. We’ve developed similar techniques in the lab to calculate the friction, and hence the car’s limits, by using measurements from the electric power steering system. The best part is that we can sense the limits of the tires when we are only halfway to them, giving us plenty of time to make corrections.



Chris Gerdes: The future race car — 150 mph and no driver


You have both a PhD in mechanical engineering and a passion for racing, so it seems like you’ve ended up in the right place. How did you come to do what you’re doing now? Can you give me a bit of your backstory?


In my undergraduate engineering classes at UPenn, I was fascinated by dynamics and control — the idea that you could describe how something moves mathematically and then calculate how to make it move differently. Wanting to learn more, I got a teaching fellowship to stay for my Masters and discovered in the process that I loved teaching even more than dynamics. I applied to PhD programs and got a call one evening from Prof. Karl Hedrick at Berkeley, who said he was looking for a “kick the tires sort of guy” for a project on automated highways. Head to California and get a PhD working with cars? Perfect.


Berkeley was fantastic for me. I minored in math to boost those skills and worked on algorithms for allowing cars to drive safely about two meters away from each other on the highway. To find a good test site, we had to head to San Diego, giving me my first taste of field testing. The project went very well and we pulled off the experiments successfully, but since I had never worked in industry, I was left wondering if what I knew applied to that elusive “real world”.


So I took a job with Daimler-Benz, the parent company of Mercedes and Freightliner heavy trucks, which took me to Germany and Portland, Oregon. I developed models of heavy trucks that could be used to test safety and performance early in the design process and helped engineers diagnose a number of strange dynamic problems. We got to work with everything from school buses to airport rescue vehicles, in simulation and on the track. I loved the work but missed the teaching that had motivated me to get the PhD in the first place.


One day a friend encouraged me to apply for a faculty opening at his university. Since I loved my job and had nothing to lose, I sat down one weekend and wrote a description of what I would do if I could do anything. I sent this off to a few universities (without much regard for exactly what the advertisements said they wanted) and Stanford took me up on the offer. I’ve been here since 1998 and feel I still learn new things from my students and faculty colleagues every day.


Out of curiosity… many individuals involved in racing are professed adrenaline junkies. Are you?


I actually associate racing more with intense focus than I do with adrenaline. I generally find myself pretty calm behind the wheel, but also totally present in the moment. The rest of the world fades away and it is just me, feeling the limits of the car or kart as I guide it around the track, trying to figure out the best way to take the position from the person in front of me. For me, the total engagement associated with racing is a way to enter the flow state. That’s pretty addictive even without the adrenaline.


What’s one problem you’re currently experiencing or an issue you’ve had to overcome during your research?


Probably the greatest challenge for us is keeping everybody on the same page. We work very closely with automotive companies, since we want our work to influence future generations of cars. Our greatest results seem to come when we get the students in our lab together with researchers in industry to develop new solutions to important problems. That’s a challenge, though, since everyone involved has different goals — universities want to generate new knowledge and publications, students are engaged in the work as part of their education, and companies are hoping to use the results as a springboard for future products. Sometimes it seems that it is harder to figure out what stickers need to go on the car than it is to do the actual research. We are fortunate that we have several industry partners with whom we have worked for years that are committed to working through these differences with us.


It seems that the future of the autonomous car is finally here, or on the cusp of here. Make a prediction for 2040. What can the next generation expect from the auto industry?


I think that by today’s standards, the cars of 2040 will look shockingly lightweight. With the ability to actively avoid accidents, either autonomously or in cooperation with the driver, car crashes will become extremely rare events. Since there is no longer a need to move all of that steel around, cars will use only a small fraction of the energy consumed today. And they will make good use of that energy — with such light weight, acceleration and handling will be excellent!


Where could I find you on a non-work day? Or rather, what inspires you — day-to-day — besides your career?


If I’m not working, I am usually hanging out with my wife, our two boys, and Sheila the Australian Shepherd. You’ll find us throwing Frisbees for the dog, out on a hike, doing projects with Cub Scouts, or continuing on our quest for the ultimate surround sound experience for family movie night.


Finally, do you have any advice for our readers? Are you looking for anything in particular right now?


In terms of advice…


I always ask my students, “Is this the coolest thing that we could be doing?” Life is short, and I think that as researchers, it is important that we are always very intentional about the problems we choose to work on.


In terms of what I am looking for…


Like anyone who races, I am always looking for ways to shave a few tenths off of our lap times!



Share your everyday innovation and you could win a $50,000 grant and the most innovative Nissan Altima ever. We’ll even unleash the winning ideas on Kickstarter. After that, anything can happen. Get started by entering your idea now at




Chris is looking for:
innovative thinkers

Stanford University, School of Engineering
Say something >

Copyright © 2012 Daily BR!NK. All rights reserved.