On The Path To Pike’s Peak: New Video of Stanford’s Robot Car

9 18 Loading
stanford robot car to race pike's peak

Shelley knocks up clouds of dust as it takes tight turns in San Jose.

It will be one of the more daring feats of robotic driving ever attempted: Stanford’s autonomous car, an Audi named Shelley, will race up Pike’s Peak at breakneck speeds. That mountain has been the proving grounds for racers since 1916, with annual competitions to see who can scale its treacherous curves in the shortest time. While not scheduled to make the Peak attempt until September, Shelley has been undergoing thorough testing. It reached speeds of 130 mph at the Bonneville Salt Flats, and ran a small tight-turn loop for the media in San Jose. All without the need for a human driver. We’ve got some great new footage of Shelley laying down some tracks as it gets tested and improved for its upcoming climb. Check out the video below.

Stanford has consistently been one of the top research teams in autonomous automobiles. They’ve grabbed first and second place in the DARPA Grand Challenge with different vehicles in different years. The Center for Automotive Research at Stanford (CARS) is also working with Volkswagen to produce a self-parking car. Racing Shelley up Pike’s peak is a new kind of endeavor, though. This will push the limits of control to get the car to its utmost performance. As graduate researcher Rami Hindiyeh explains in the video, they’re looking to get the maximum force on Shelley’s tires at all times:

At these speeds and turns, CARS will push Shelley as close to failure as they can. Which could have some serious applications if robot vehicles are ever to take over for human drivers. Being able to recover from dangerous turns and rapid accelerations will make computer-driven cars safer. Just as automatic systems reroute brake pressure and torque on modern cars, so too may we see systems in place that will augment steering controls in the near future.

Seeing robot cars on the highway, however, is still quite a ways off. Shelley uses GPS to navigate on the curviest roads, but I haven’t seen anything that shows me it could brake in time to avoid a pedestrian, or dodge road debris. A highway, with dozens of moving variables to account for, is still beyond the capabilities of autonomous vehicles. Give it time though, and I’m sure we’ll see Stanford, or another team, develop the means to overcome that challenge as well.

For now, all eyes are on Pike’s Peak. CARS is continuing its research, and gathering financial momentum as well. State Farm Insurance signed on as an affiliate, the sixth major company to do so. There’s obviously a lot of interest in projects like these, which may demonstrate that computers already have superior car-handling skills, if not better judgment, compared to the average driver. Humans will race up Pike’s Peak for the annual competition in late June, setting the bar for true rally driving performance. Will Shelley run the course at competitive speeds in September? We’ll see.

[screen capture and video credit: Jack Hubbard via Stanford News]
[source: CARS, Stanford News]

Discussion — 18 Responses

  • Don April 14, 2010 on 3:06 pm

    Oh no, is this click-through-for-the-content stuff for real?

  • Don April 14, 2010 on 11:06 am

    Oh no, is this click-through-for-the-content stuff for real?

  • Fletch April 14, 2010 on 5:02 pm

    To really push home to viewers that there is NO DRIVER in the car, they need to completely white out all the windows of the car during the race. This race is more than just a test, it’s a statement to humanity that the singularity is coming. Watching a unmanned car race up a mountain aggressively and possibly better than the other drivers will raise a red flag with the public. It will demand their attention.

  • Fletch April 14, 2010 on 1:02 pm

    To really push home to viewers that there is NO DRIVER in the car, they need to completely white out all the windows of the car during the race. This race is more than just a test, it’s a statement to humanity that the singularity is coming. Watching a unmanned car race up a mountain aggressively and possibly better than the other drivers will raise a red flag with the public. It will demand their attention.

  • John Gehrke April 14, 2010 on 6:37 pm

    I agree with the idea of painting the windows, but I could still wonder about the possibility of human remote control.

  • John Gehrke April 14, 2010 on 2:37 pm

    I agree with the idea of painting the windows, but I could still wonder about the possibility of human remote control.

  • Ben April 15, 2010 on 8:57 am

    Sod all that. I just want to see how quickly it could get round the Top Gear test track.

    • RagnarDanneskjöld Ben November 12, 2010 on 3:34 pm

      Oh Gawd, Imagine the Robo-Stig as software…

  • Ben April 15, 2010 on 4:57 am

    Sod all that. I just want to see how quickly it could get round the Top Gear test track.

  • Matt April 26, 2010 on 2:23 pm

    I would like to point out that there is an error in the article about the autonomous cars only using GPS. Stanford’s previous cars (and I am almost certain Shelly) all use Ladar and other object recognition devices to avoid collisions and dodge obstacles. Additionally previous cars were driven next to professional drivers, demonstrating their ability to recognize and respond to other moving vehicles. That being said the concept is very exciting, but it’s primary obstacle to actual on road (with real people) use is not response capability, but more assurance of absolute reliability (people fear autonomous control) and cost.

  • Matt April 26, 2010 on 10:23 am

    I would like to point out that there is an error in the article about the autonomous cars only using GPS. Stanford’s previous cars (and I am almost certain Shelly) all use Ladar and other object recognition devices to avoid collisions and dodge obstacles. Additionally previous cars were driven next to professional drivers, demonstrating their ability to recognize and respond to other moving vehicles. That being said the concept is very exciting, but it’s primary obstacle to actual on road (with real people) use is not response capability, but more assurance of absolute reliability (people fear autonomous control) and cost.

  • Nastyman May 11, 2010 on 12:51 pm

    So you can’t get a OUI in this then right?

  • Nastyman May 11, 2010 on 8:51 am

    So you can’t get a OUI in this then right?

  • Chris May 13, 2010 on 1:55 pm

    Already been done. The cogged train goes up to Pikes Peak faster and more directly than the road up — and more efficiently. This isn’t a race to say a robot/machine can do the ascent better. It is a leap forward in algorithm design and input to say a robot can handle varying real world situations.

    If you want optimized robot cars just lay down train tracks.

  • Chris May 13, 2010 on 9:55 am

    Already been done. The cogged train goes up to Pikes Peak faster and more directly than the road up — and more efficiently. This isn’t a race to say a robot/machine can do the ascent better. It is a leap forward in algorithm design and input to say a robot can handle varying real world situations.

    If you want optimized robot cars just lay down train tracks.

  • jim May 13, 2010 on 6:29 pm

    I agree. This is Hollywood grade, not problem solving grade.

    Do something useful. Contribute to wikiSpeedia to help robot-cars of the real-future.

  • jim May 13, 2010 on 2:29 pm

    I agree. This is Hollywood grade, not problem solving grade.

    Do something useful. Contribute to wikiSpeedia to help robot-cars of the real-future.

  • dvd gps September 27, 2010 on 3:39 pm

    I agree with the idea of painting the windows.