A Google patent application made public today goes into great detail about the possibilities of its self-driving car project — and no, this isn’t an April Fool’s joke.

According to the filing, features would include the ability to autonomously pick up kids from school and drive them home, communicate with other vehicles to avoid obstacles such as a deer in the road, let people doze on special “sleeping surfaces” inside the car while the computer drives, and change shape automatically depending how the vehicle is being used, among many other features.

The filing also notes that the technology could be applied to a variety of different vehicles, not just cars and trucks but “busses, boats, airplanes, helicopters, lawnmowers, recreational vehicles, amusement park vehicles, trams, golf carts, trains, and trolleys.”

Google first detailed the autonomous car project in 2010, using technology including video cameras, radar sensors and a laser range finder to allow the cars to drive themselves. Prototypes have alternatively thrilled, impressed and scared earlier testers, as shown in the video above. The project is an outgrowth of U.S. DARPA Challenge for autonomous vehicles.

The patent filing was submitted in October and made public by the U.S. Patent & Trademark Office today.

Combined with Google’s Project Glass, an augmented reality glasses initiative made public this week, the patent document fills in key details about Google’s vision for the future role of computers and online services in our lives.

Another detail from the patent filing: The cars could come with personalized settings to reflect the user’s driving style.

The filing explains, “For example, a more aggressive driver may want to change lanes more often to pass cars, drive in the left lane on a highway, maneuver the vehicle closer to the surrounding vehicles, and drive faster than less aggressive drivers. A less aggressive driver may prefer for the vehicle to take more conservative actions, such as somewhat at or below the speed limit, avoiding congested highways, or avoiding populated areas in order to increase the level of safety.”

Other possibilities include “autonomous law enforcement vehicles such as patrol vehicles, fire trucks, or ambulances” that could be given limited control of nearby vehicles — ensuring, for example, that people pull over when they pass.

“During dangerous high-speed chases, the vehicle may allow autonomous law enforcement vehicles to maintain high speeds in confined situations, such as through areas of high traffic, while at the same time reducing the likelihood of a collision with environmental objects,” the patent filing says.

The company recently showed how the car was able to drive a man who is 95 percent blind.

Like what you're reading? Subscribe to GeekWire's free newsletters to catch every headline


  • Guest

    Now you just need MS’s Craig Mundie to comment about how MSR has been doing pretty much the same thing only years earlier. ;-)

    I wonder what insurance company would insure a self-driving car? Although where I live, I can’t imagine a computer driven car being any worse than the average driver on the road currently.

  • Guest

    Congratulations to Google on continuing to innovate!

  • Dracono

    Very cool demo, but I’ll file this next to Utopia for everyday use. For now this will be more practical for confined areas or industrial use, say something like farming supervised with a kill switch of course.

  • http://www.christopherbudd.com Christopher Budd

    When I was with the Microsoft Security Response Center I always said “I don’t want to ever patch my car”.

    I still stand by that.

    All due respect to Google (and anyone else) but I’ve seen too many “oh crap, I didn’t think of that” moments from engineers to feel comfortable about putting my life in their coding hands like this.

    And making the systems networked is really risky from a security point of view. You can’t attack and endpoint system you can’t reach and that makes it reachable.
    Besides, like I always said, this raises the stakes around the seriousness when the system “crashes”.

    • http://www.facebook.com/matjnewton Matthew Newton

      Christopher I doubt those moments would be as bad as the “Oh, Crap” moments that kill 30,000 people every single year on our roads.

      The engineers working on this are the best in the world which means they know the exact issue that you have raised – so instead of trying to “think” about situations they are just out there doing… and have clocked well over 200,000 miles in the field as a result. Smart cookies, these guys. They’re two steps ahead!

      Making the systems networked is not necessary for driverless cars to operate, just as it is not necessary for cars to operate today.

      Matthew Newton
      Driverless Car HQ

      • http://www.christopherbudd.com Christopher Budd

        Thanks for the reply. I’m sure you’ve got smart people working on it, but one major difference between computers and humans is that computers are a closed system, they can only take action based on their programming and that programming has to think of all possible needs in advance.

        Humans can adapt and improvise.

        Yes other drivers have plenty of oh crap moments, but I as a human driver can improvise a response as needed. I’m not comfortable being at the mercy of a closed system where my physical safety is concerned.

        I’d encourage you and your team to read about the Air France 447 crash and the role of autopilot in that as a cautionary tale:


  • http://www.davidkesslerauthor.com/ David Kessler

    I’ve just looked through the claims and couldn’t see anything that qualifies as an inventive step or a NON-obvious extension of prior art.

Job Listings on GeekWork

Find more jobs on GeekWork. Employers, post a job here.
Software Engineer – “Test Guru”Allen Institute for Artificial Intelligence