Stanford’s Self-Driving Car Tears It Up On Racetrack – Tops 120 MPH

27,370 6 Loading

Shelley, Stanford's self-driving car, tops 120 mph on racetrack straightaways, but still has work to do to catch up to human drivers.

Just as Google’s self-driving Prius goes for distance, recently passing 300,000 miles, Stanford’s self-driving Audi TTS instead has the need for speed. The Audi, known as Shelley, sped around the Thunderhill Raceway track north of Sacramento topping 120 miles per hour on straightaways. The less than two and a half minutes it took to complete the 3-mile course is comparable to times achieved by professional drivers.

Shelley is the product of a collaboration between Stanford’s Dynamic Designs Lab and the Volkswagen Electronics Research lab. In 2010 Shelley put the pedal to the medal, speeding and drifting through the 156 turns up to the top of Pike’s Peak racecourse. The current run through Thunderhill is another example of how Stanford and Volkswagen engineers are set on pushing robotic car performance to its limits, quite different from Google’s singular goal of teaching their car how to obey traffic signs and avoid joggers. Sure Google’s will most likely be the first self-driving cars to legally take passengers, but they still can’t slide into parking spaces James Bond style.

Shelley’s fast, but not quite as fast as humans. Professional driver times at Thunderhill still beat Shelley’s times by a few seconds. Associate Professor Chris Gerdes, who heads the Dynamic Designs Lab at Stanford, is trying to close the gap analyzing runs at the track and using the data to tweak the car’s software. Humans outperform Shelley because they’re guided by feel and intuition and their paths through the track are very smooth. The algorithms that guide Shelley’s driving involves continual course corrections. In an effort to catch up to humans, Gerdes is trying to develop algorithms that make Shelley more humanlike. “Human drivers are OK with the car operating in a comfortable range of states,” Gerdes said in a press release. “We’re trying to capture some of that spirit.”

To quantify that ‘spirit,’ Gerdes and his team will place physiological sensors on two human drivers at the Rolex Monterey Motorsports Reunion races this month. The sensors will measure, among other things, their brain activity with scalp electrodes in an attempt to pinpoint times of heightened concentration. Assumedly it is during those times that human calculation deviates most from software algorithm.

It’s only a matter of time before robotic cars outperform their human-driven counterparts. But the team’s ultimate goal is not to create yet another robot who can best their human competitors. Extreme speeds on a racetrack can sometimes simulate perilous conditions it might encounter on the normal road. The algorithm it needs for a spinning wheel to regain traction could be very similar to straightening out after sliding on an icy surface. And as we all know, even everyday driving in fine weather can suddenly turn into a death-defying act of avoidance and control. Lessons learned on the racetrack could better prepare Shelley and her successors for those split-second reactions. Who knows, maybe the high-speed tactics could be incorporated into fighter drones like the Navy's X-47B to be a more efficient hunter-seeker.

The road ahead for Shelley and Google’s self-driving car is still a long one, and both seem to be taking different routes to providing the world with a safe self-driving car. In the time being, watch Shelley tear it up in the following video and see if you don’t have the same question I do: When’s it my turn?

[image credits: Stanford University via YouTube]
video credit: Stanford University via YouTube
images: Stanford University
video: Stanford University

Peter Murray

Peter Murray was born in Boston in 1973. He earned a PhD in neuroscience at the University of Maryland, Baltimore studying gene expression in the neocortex. Following his dissertation work he spent three years as a post-doctoral fellow at the same university studying brain mechanisms of pain and motor control. He completed a collection of short stories in 2010 and has been writing for Singularity Hub since March 2011.

Discussion — 6 Responses

  • Ivan Malagurski August 19, 2012 on 12:01 pm

    Way to go Stanford!

  • arpad August 20, 2012 on 3:36 am

    That’s a neat demo although once you can go slow on a road with no surprises on it you can also go fast. What concerns me about autonomous cars is the ability to handle all the myriad unanticipated, and in many cases unanticipatable, dangers that can rear up suddenly.

    For instance, can an autonomous car detect and compensate for or avoid black ice? The laser range-finder just went out. What’s the car do? The car just went from half mile visibility fog to twenty yard visibilty fog. What now? A cat darts out into the street. What’s the car do? A cat darts out into the street from the right as an incautious driver opens his door on the left. Go straight and hit the cat while avoiding the drivier? Swerve and avoid the cat hitting the driver?

    The core problem of autonomous cars seems to have been solved. Now the edge conditions have to be addressed and it’ll be interesting to see how the smart folks at Standord and other schools adress the problem.

  • Improbus August 20, 2012 on 10:46 am

    I can’t wait until Google can enter one of these cars in a NASCAR race … and win.

  • Jessica231 August 20, 2012 on 4:46 pm


  • mansky August 21, 2012 on 1:50 pm

    are you going for the pun, or did you mean “pedal to the metal”? (petal to the mettle?)

  • Cynic January 26, 2013 on 12:34 am

    i want one