Forgot your password?
typodupeerror
Robotics Transportation Hardware Technology

Stanford Robot Car Capable of Slide Parking 265

Posted by samzenpus
from the no-human-would-drift-like-that dept.
kkleiner writes "Stanford's Junior, the robot car that took second place at DARPA's Grand Challenge in 2007, has learned how to perform a tire-squealing 180-degree spin into a skin-tight parking space. Similar to a James Bond action scene, the maneuver is impressive and would be extremely difficult for a human to pull off. We won't be handing the keys over to robot cars anytime soon, but Stanford shows us that at least for some driving tasks robot cars can already meet or even exceed human ability."

This discussion has been archived. No new comments can be posted.

Stanford Robot Car Capable of Slide Parking

Comments Filter:
  • Rude-bot (Score:4, Funny)

    by Tablizer (95088) on Wednesday May 12, 2010 @10:12PM (#32188888) Homepage Journal

    To counter, I'm inventing the Automatic Finger to quickly signal my frustration at being cut-off from my parking spot.

  • Toyota (Score:5, Funny)

    by phantomfive (622387) on Wednesday May 12, 2010 @10:21PM (#32188930) Journal

    We won't be handing the keys over to robot cars anytime soon

    Heh.....let's work on getting cars to stop reliably before we start talking about that

  • Wind noise (Score:3, Funny)

    by Itninja (937614) on Wednesday May 12, 2010 @10:28PM (#32188958) Homepage
    I know it's all bobo-chis to shot amateur video and whatnot, but can a brother get some noise reduction up in here? I mean, DAMN.
    • by Barny (103770)

      Well, they did have a Metallica song as the backing, but, well, Lars nearly had an aneurysm that something cooler than him was using their music as a backing...

  • by bziman (223162) on Wednesday May 12, 2010 @10:36PM (#32189014) Homepage Journal
    Do notice all the tire tracks that don't lead into the parking spot? Like all of the robotics projects I've been involved in, this took a LOT of tries.
    • by shutdown -p now (807394) on Wednesday May 12, 2010 @11:13PM (#32189212) Journal

      IF you read TFA (a novel concept, I know!), it has a longer video which demos several different algorithms which fail variously; and then, ultimately, a final run which combines all of them to succeed. They claim that it is this smoothless combination is what is the real innovation here.

      • by TheLink (130905)
        They could easily just show the video where the "combined algorithm" succeeds, even if there were times where it failed.

        Anyone seen the endings of those Jackie Chan movies? In the movie proper you see the fancy successes. The ending is where you see some of the other clips where stuff went "different".

        Since they have resorted to "open loop" control, it means the robot is about as likely to make as many mistakes as humans. It's progress I guess, but really, if it can only get within 2 feet of actual spot, I
      • Can I have said Robot Transform into Shopper With Recyclable
        Grocery Bag and then pick up my groceries?

        WHat do you say, Stanford? (Forget about the Toss-keys-to-valet
        trick, we KNOW you guys can make it do that, eventually. We're
        damn impressed by this one already!)

    • If you read the article, you will see that they tried three different algorithms. The first two failed, but the final one succeeded quite reliably.

      So it's not like they were trying a bunch of times and then only filmed the one that worked (I assume that's what you meant, your post wasn't quite clear).
    • by pz (113803)

      Do notice all the tire tracks that don't lead into the parking spot? Like all of the robotics projects I've been involved in, this took a LOT of tries.

      Sure, sure, but did you watch the rest of the videos where the demonstrate how single-modal control fails badly? That's where the other tire marks are from.

      The point of the set of videos was not that the car can do slide parking per se, however incredibly awesome that is, but that it took a multi-modal approach to achieve it.

  • Autonomous? (Score:3, Insightful)

    by Itninja (937614) on Wednesday May 12, 2010 @10:36PM (#32189020) Homepage
    If a robot does exactly what it is programmed to do, is that autonomy? Is sounds like they programmed it to perform this maneuver (going backward at a decent speed and sliding into a specific zone), and then it did (after several failures apparently).
    • Yeah, but programming it to do it is real hard. Yeah, it's for damn sure that a system like this isn't ready for prime time, but it's still an accomplishment.
      • by Itninja (937614)
        It is quite a programming feat to be sure, but the difficulty of the work really has nothing to do with the autonomy of whatever product said work produces.
  • by gillbates (106458) on Wednesday May 12, 2010 @10:49PM (#32189090) Homepage Journal
    1. Guided missiles have been able to direct a few hundred pounds of ordnance traveling at mach 5 into a spot the size of a trash can for the last few decades now. The fact that a computer can park a car with perfect precision - while cool - is hardly revolutionary.
    2. Liability and lawyers have done more to keep advancements like these from the market than any engineering challenge every has. The first time a car parallel parks into a toddler, you can kiss the feature goodbye.
    3. While I don't intend any sexism, the fact is that parallel parking is difficult, and many women would rather have the car do it than try it themselves. Parking assist (without the tire-scorching 180) is going to be a very popular feature among the minivan crowd.
    4. This is *really bad* from a freedom perspective. As soon as computers can reliably and safely drive cars, anyone who *dares* drive the car themselves will be considered negligent; they'll probably pass a law against it. And with pervasive networking, you can be assured the police will have the ability to remotely disable your vehicle, should the need arise (the can already do it with GM vehicles, which is problematic enough...). I do not look forward to a future where my movement *in my own vehicle* can be arbitrarily and capriciously monitored and regulated remotely.
    • Re: (Score:2, Insightful)

      by Anonymous Coward

      While I don't intend any sexism ...

      Then why did you make any reference to gender at all? You could have made the exact same point without introducing any stereotypes.

    • by trawg (308495) on Wednesday May 12, 2010 @11:11PM (#32189194) Homepage

      While I don't intend any sexism, the fact is that parallel parking is difficult, and many women would rather have the car do it than try it themselves. Parking assist (without the tire-scorching 180) is going to be a very popular feature among the minivan crowd.

      heh every girl I know is awesome at parallel parking. I (male) have done it a grand total of twice in the more than ten years since I got my license; I would rather drive another few metres and walk a bit further than embarrass myself trying and hold up traffic.

      I know it's the sort of thing I would learn with practice, but I drive infrequently and its just not a skill I need where I live, so I just can't do it.

      For me, driving is a painful chore that I must do to get from point A to point B. I know some people enjoy it but I find it tiresome, boring, and a waste of time. I cannot wait for the day that I can get in my car and punch in a destination and sit back and read a book or idly stare out the window!

    • by Sibko (1036168) on Wednesday May 12, 2010 @11:16PM (#32189232)
      While I agree there are some issues with freedom therein, I can also see how taking control of motor vehicles out of the hands of the general person, and into the hands of a computer is going to massively improve safety.

      Think of all the dangerous maneuvers and careless driving you've seen. That will be no more - speeds on highways can be increased, distances between cars can be decreased. You won't be driving the car yourself anymore, but you'll get to your destination faster and more safely than you do now.

      Anyways, I very much doubt manually driving a vehicle will ever disappear. Driving is fun, it's something people enjoy - what this will simply do is eliminate all those times you wouldn't have fun driving, by giving control over to the computer whenever you feel like stopping.
      • We don't need fancy robots; we need better driver training. In the US, you demonstrate basic proficiency in skills that matter 95% of the time when everything is going swimmingly, answer a very limited subset of the rules/laws of the road, and then get handed your license, and never need to do any of that again. Why are we shocked when people then miserably fail when the shit hits the fan? In other countries, you have to learn and demonstrate actual car handling skills, like recovering from a skid...and

        • Re: (Score:3, Insightful)

          We don't need fancy robots; we need better driver training

          Yes Comrade, and as soon as we create the New Soviet Citizen, communism will work.

          Anytime your plans call for people suddenly and magically becoming smarter or more responsible, you have an issue.

    • by MojoRilla (591502) on Wednesday May 12, 2010 @11:31PM (#32189304)
      I'd gladly hand control of my car over to a computer if it would get me to work in half the time, and let me do other things while I "drive". And this will someday happen. Where I live, traffic slows down because of a bend in the highway. Rubbernecking causes massive jams and delays. There has to be a better way.
      • by chromas (1085949)
        A series of (pneumatic) tubes.
        Get the scientists working on the tube technology, immediately. (tube technology)
        Chop chop! Let's go!
    • by mmaniaci (1200061) on Wednesday May 12, 2010 @11:40PM (#32189344)

      anyone who *dares* drive the car themselves will be considered negligent

      That's about the only thing you said I agree with. Driving is stupidly dangerous [wikipedia.org] and I can't wait until computers are driving every car on the road. Freedom? Pfff, driving a car doesn't make you free. The idea is just marketing done by the auto industry to make you feel American and Free by buying 5 cars per family and the gas that goes with 'em.

      Also, there is a HUGE FUCKING DIFFERENCE between a bomb, where the payload is intended to destroy, and a car carrying passengers. Think about fault tolerance...

      • by e4g4 (533831)

        Think about fault tolerance...

        We all know the value of splash damage...

        But in all seriousness, driving with the intent of getting yourself to a destination is no fun at all. There are all these rules, speed limits, and other drivers ruining the whole experience. Cars as a mode of personal transportation are horribly inefficient and unbelievably dangerous. The act of driving itself can be tons of fun - if you enjoy driving, do it on a track (at whatever speed, and performing whatever maneuvers you'd like). The rest of us will take

        • Re: (Score:3, Interesting)

          by adolf (21054)

          Driving in circles on a track is also no fun at all if you want to, you know, eventually end up someplace else.

          But I think I have a good compromise:

          You keep riding the subway in $bigcity, and I'll keep driving my cars in rural Ohio for fun. These are not mutually-exclusive things.

      • Re: (Score:3, Insightful)

        by istartedi (132515)

        What's always interesting about those statistics is how big a difference recessions make (fewer people rushing to work, fewer accidents) and how little difference speed limits make.

    • by Chuck Chunder (21021) on Thursday May 13, 2010 @12:01AM (#32189422) Homepage Journal

      While I don't intend any sexism

      I guess it must just come naturally then.

    • This is *really bad* from a freedom perspective. As soon as computers can reliably and safely drive cars, anyone who *dares* drive the car themselves will be considered negligent; they'll probably pass a law against it.

      Probably only on public (taxpayer-funded) roads. Which I'm entirely in favor of. Public infrastructure exists to meet the needs of the public at large, not for individuals to play at being Dale Junior. I'm sure "stunt driving" by humans will remain legal on closed courses and tracks.

      I expect a transitionary period with controlled-access roads like interstate highways and motorways going automation-only first, and surface streets following. And frankly, I can't wait.

    • This is *really bad* from a freedom perspective. As soon as computers can reliably and safely drive cars, anyone who *dares* drive the car themselves will be considered negligent; they'll probably pass a law against it. And with pervasive networking, you can be assured the police will have the ability to remotely disable your vehicle, should the need arise (the can already do it with GM vehicles, which is problematic enough...). I do not look forward to a future where my movement *in my own vehicle* can be arbitrarily and capriciously monitored and regulated remotely.

      You know, there's a reason why "slippery slope" is classified as a logical fallacy rather than a logical argument.

    • Re: (Score:3, Funny)

      by mgblst (80109)

      I know, how dare anyone post something that isn't COMPLETELY REVOLUTIONARY, and change the way everybody does everything, in their lives. Anything less than that is not worthy of my attention. That is why I am so much fun at parties.

    • Re: (Score:2, Insightful)

      by sl149q (1537343)

      Look at what the intense lobbying and marketing going into anti-drunk driving. The best example being MADD (Mothers Against Drunk Drivers.) To (rightly) save +10,000 people a year from being killed.

      Once we have robotic controlled cars that can reduce traffic accidents by a suitable amount there will be a similar incentive to get rid of the bad driving of humans. Think MAHD (Mothers Against Human Drivers.) And I think saving another 10,000 plus people a year and further reducing insurance rates etc will make

    • Re: (Score:3, Insightful)

      by Krneki (1192201)
      Leave the AI to drive your car and if you feel bored play the last car simulation while you wait to arrive at your destination.

      There is no reason we have to wait stuck in traffic because you feel you are losing your freedom to drive like a jackass. :)
    • Re: (Score:3, Insightful)

      by otter42 (190544)

      1) Yeah, that's just wrong, and missing the point. We can guide missiles into tiny spots because we have incredibly good models of their flight path. We can drive a car into a tiny spot in exactly the same way. What's interesting here is that they mix together a LQR controller with open-loop, in a way that does not require hand-tuning and gives excellent, repeatable results.

      If we had a dynamic model of the car as it were sliding sideways, I'm sure we could use Lie brackets to discover all sorts of interesti

  • Why not just put mecanum wheels on an all wheel drive electric car?
    • by yukk (638002)

      Why not just put mecanum wheels on an all wheel drive electric car?

      ... Because they want it to go more than 5km/h ?

  • by SuperBanana (662181) on Wednesday May 12, 2010 @11:22PM (#32189256)

    Similar to a James Bond action scene, the maneuver is impressive and would be extremely difficult for a human to pull off.

    Bullshit [youtube.com]. Yeah, he's one of the best- but he's doing that in a 500hp AWD car, not a 100HP FWD diesel station wagon, at speeds several times higher than what Stanford was doing. Call me when they can do what he does.

    It's also extremely difficult for a human to pull off crochet if they haven't been taught how. Or to shoot a rifle and hit a target a mile away. Or fly fighter jets in formation feet apart. Yet we do that. The question is: how hard is it to train someone, and how consistently can they do it, and how much effort did it take to get the computer to do it?

    The answer to the first part: Top Gear did a show segment where they had Russ Swift [youtube.com] teach a bunch of people off the street how to do it. If I recall, they were grandmothers. They were going for a larger area, but come on- they were octogenarians.

    The answer to the second part:

    Apparently Stanford hasn't heard of rallying or gymkhana. Tens of thousands of people do stuff way, way more impressive than what Stanford is demonstrating- at much higher speeds in much more powerful cars. It's not hard, and the Stanford guys are grossly overexaggerating the complexity of the problem to model, as well. The whole point is that you use the car's momentum and lock wheels to make it slide predictably. Practice makes perfect for timing and aim (in the case of Top Gear, they practiced with inflatable boxes that were harmless to the cars.)

    And, how many tries do you suppose it took the Stanford team to get it right?

    • by pclminion (145572)

      Wow. What bug crawled up your ass? If it's so fucking easy, why the hell should I be impressed by what's-his-face doing it?

    • by HBoar (1642149) on Thursday May 13, 2010 @12:13AM (#32189480)

      Yeah, he's one of the best- but he's doing that in a 500hp AWD car, not a 100HP FWD diesel station wagon, at speeds several times higher than what Stanford was doing.

      It's a hell of a lot easier to pull off maneuvers like that in a powerful, lightweight, AWD car set up for sliding than it is in a family wagon....

      Do you know anything about control theory? At all? It certainly doesn't sound like it. They're not just programming the car what to do and when, the car sees where the cones are and works it out for itself.

    • Re: (Score:3, Funny)

      by mgblst (80109)

      Call me when they can do what he does.

      Ok mr super unimpressed, you forgot to leave your phone number.

    • I'll raise you one: Ari Vatanen driving up Pikes Peak:
      http://www.youtube.com/watch?v=TKgeCQGu_ug [youtube.com]

      And Ari's not only a wonderful driver, he doesn't need to do spin-out videos just to show off how great he is.

    • by wvmarle (1070040)

      Cool video, cool stuff. But there is good reason these stunts are done on closed tracks only.

      Anyway to come back on learning capacity: we have been building computers for just over half a century now, and they are getting quite powerful. And indeed some AI is appearing.

      OTOH the human brain has been in development for millions of years. Admittedly trial and error is less efficient than the direct approach like we have with computers, still I would say the learning capacity of our brains is far more mature

    • Re: (Score:3, Insightful)

      by nedlohs (1335013)

      Wow, way to miss point and be an idiot at the same time.

      You must get really pissed off when people show off their computer programs that do natural language processing. I mean a 6 year old do better, and we manage to teach almost anyone to do it without difficulty much better than the stupid computer can.

      Or their little robots doing stuff with video recognition in order to recognize and move blocks around - I mean toddlers can do that!

  • by NotQuiteReal (608241) on Wednesday May 12, 2010 @11:34PM (#32189322) Journal
    If one parks a car this way, is it possible to un-park it?
  • Meh (Score:5, Insightful)

    by BluBrick (1924) <blubrick@@@gmail...com> on Wednesday May 12, 2010 @11:36PM (#32189330) Homepage
    I'll get the car into the tight spot - call me when the car can autonomously find a parking spot.
  • by Eil (82413) on Wednesday May 12, 2010 @11:42PM (#32189346) Homepage Journal

    But where's the style? I thought it was common knowledge that the very first thing you're supposed to do after you've programmed a park to screech into a parking space is install a loudspeaker behind the grille which yells out, "heeeee-like a glove!"

    • by Eil (82413)

      goddamn it, I'll sooner win the lottery than post a joke to slashdot without a glaring typo in it

  • by 1984 (56406) on Wednesday May 12, 2010 @11:56PM (#32189400)

    ... before it even happened. A few Lexus introduced the automatic parallel parking feature, and Audi responded with this:

          http://video.google.com/videoplay?docid=3593724097279407250# [google.com]

    Amusing retort. Irrelevant for 99.9%+ of people, but sold right into the person you'd love to be.

  • This really isn't so surprising at all. Humans have a rather limited sensor suite compared to some of the robotic systems we can now build. Maneuvers like the one described are all about full panoramic visibility and awareness, something humans don't really possess... at least not individually. Is it any wonder that humans so often cooperate in similar situations where one set of fixed eyes and ears really isn't enough? Think about a squad of soldiers: it's as much about combined awareness as it is comb

  • So basically they show that their car can learn by copying some known process in addition to their physics model.

    Yet I'm curious to know how the car decide that they should use the learned process rather than their model. I mean, if they learned this parking maneuver on gravel, and tried that on slick pavement, anybody would probably fail. Can this car actually do better? I would be curious to see if the car can learn a maneuver in some given conditions and repeat it in slightly different conditions.

    • by pipedwho (1174327)

      In theory, the AI could learn (and/or be taught) just about every conceivable road surface and driving condition and how the parameters are inter-related. With appropriate external sensors, and for a single test manoeuvre, it is likely to be able to best just about any human driver.

      This is simply an example of a robotic driver AI being demonstrated with more flare than the usual run-of-the-mill reverse park.

      A vehicle AI driving blindly is equivalent to letting a blind man drive a car. The real innovation co

  • Standard on the Lexus LS460L, though not as thrilling. The Japanese auto manufacturers just don't seem to have a flare for the dramatic.
  • You should see it play GTA [wikipedia.org], dawg.
  • Various members of the Stanford DARPA challenge team have spoken at the IEEE Robotics & Automation society meetings here in Sili Valley, including Sebastian Thrun. All said at one point or another that one of their secrets of success was tuning the driving algorithms to "drive like Sebastian" (direct quote) because he "is an aggressive driver" (direct quote, delivered with a tone of understatement.) Sebastian Thrun likes cars... and robots. Right now I'm wondering what his arrival in the faculty park

  • So... (Score:4, Funny)

    by durrr (1316311) on Thursday May 13, 2010 @04:41AM (#32190444)
    In the future, not only will cars drive themselves, they will do so in the most action packed manner possible.
    I'm looking for funding to put ramps and pyrotechnics all over national highways.
  • by zkolter (1810670) on Thursday May 13, 2010 @06:45AM (#32191006)
    Hey everyone,
    Thanks for checking this out! To answer a few questions that have been asked:

    This video actually was indeed shot the first time we put the whole system together. Of course there were other runs, both demonstrating the alternative approaches and before we had everything working properly, that didn't succeed, but the final system was pretty reliable as autonomous driving goes. That said, we'd want to test this quite a bit more before I'd be willing to lie down where those cones are, and a big issue here is that the maneuver does shred through tires pretty quickly and is pretty tough on the car in general :-).

    Second, I certainly wouldn't argue that what we're doing here rivals the very best human drivers (the claim we're making is just that this is one of the more challenging _autonomous_ maneuvers that has been demonstrated). The best humans are certainly able to drive incredibly impressive stunts, and we only claim to be making progress towards this level of ability. However, it's worth noting that this particular maneuver is probably one that _most_ people would have trouble with (I know I certainly can't do it!).

    Let me know if there are any other questions, and I'll do my best to clarify.

    Thanks!
    Zico

    • Re: (Score:3, Insightful)

      by Toze (1668155)

      Sir;

      Thank you for building the future. It looks neat.

      Respectfully yours,
      The present.

FORTRAN is for pipe stress freaks and crystallography weenies.

Working...