Forgot your password?
typodupeerror
AI Robotics The Military

DARPA's Atlas Walking Over Randomness 76

Posted by Unknown Lamer
from the wait'll-it-finds-a-gun dept.
mikejuk writes "Considering how long we have been trying to solve the problem, a robot walking is mostly amusing. Atlas is an impressive robot, evoking the deepest fears of sci fi. Watch as one of the DARPA challenge teams makes Atlas walk, unaided, on randomness. This video of Atlas was created by the Florida Institute For Human and Machine Cognition robotics team. It shows Atlas walking across a random collection of obstacles. Notice that even though it looks as if Atlas is supported by a tether, it isn't — as proved when it falls over at the end."

This discussion has been archived.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

DARPA's Atlas Walking Over Randomness

Comments Filter:
  • ... trips at the end. The real problem is that how walking works in humans isn't understood. If it was we could simply apply the model via mimicry to bi-pedal human like robots.

    The vibration of the robots feet is probably the most jarring aspect of the video. No human has vibrating feet.

    • Re:The robot.. (Score:4, Insightful)

      by Dunbal (464142) * on Saturday November 23, 2013 @08:37PM (#45504323)
      It's understood perfectly. It's just very very hard to simulate, as there are many muscle groups involved from your toes up to your head and your shoulders, all moving in a coordinated effort. Getting it exactly right on a robot is - hard.
      • One thing I haven't seen on either this or ASIMO is a back with lordosis [wikipedia.org]. This is critical for our balance while walking, and may allow these robots to get rid of the knee bend.
      • "It's understood perfectly. It's just very very hard to simulate"

        The two things don't go together. If it was understood as perfectly as you claim you could take apart the understood model and make simple versions of it that work. i.e. if you are in possession of the fundamental principles of bi-pedalism in all environments you can then reduce complexity because you understand the fundamental concepts of walking given any set of normal human terrain.

        We haven't got anywhere near that with robotics, so to sa

        • Only if reductionism works for walking. There are some systems where simplifying the system removes the interesting behavior. You may be able to understand the complex system, but be unable to build a replica due to physical limitations.
          • To extend on blahplusplus' comment, I've heard a wonderful quote the other day: "If you can't simplify a problem, you don't really understand it." I'd need some serious convincing to accept that bipedal robots have some sort of exemption from that rule.
        • There are a bunch of things we think we know how we could do them, if only we had the [material, machines, power] to build them. We can describe what we need, but not how to make them.

          Of course, often times it seems that once we get those things, we get to move on to the next part of the problem we didn't see before.

    • Re: (Score:3, Insightful)

      by The123king (2395060)
      also, pretty much every human being would make sure to not step on any of those obstacles, mainly because they don't want to trip and fall. Who seriously sees a corridor full of wood and goes "fuck it, i'll just stroll straight over all this"? You don't, you look to make sure you're stepping on solid ground. Sure, there's times when you do have to tread on uneven and unsecured surfaces, but then you take even more care on where you're going. This robot doesn't seem to have the logic to do this, and that's w
      • by citizenr (871508)

        Sure. Robot would have to
        -look at the ground
        -understand whats on the ground, flat vs obstacles
        -understand solid vs movable/soft obstacles

        we are barely at step 2 nowadays, even for quite predictable environments like roads.

      • by fractoid (1076465)
        While I 100% agree that this problem should be solved by better cognition and route planning rather than better reflexes and balance, maybe they were working on its ability to handle uneven ground rather than its ability to avoid it? Often when developing systems like this, you have to deliberately put the system in a situation which it can't (yet) handle, in order to improve how it handles said situation.
      • by GodGell (897123)

        also, pretty much every human being would make sure to not step on any of those obstacles, mainly because they don't want to trip and fall. Who seriously sees a corridor full of wood and goes "fuck it, i'll just stroll straight over all this"? You don't, you look to make sure you're stepping on solid ground. Sure, there's times when you do have to tread on uneven and unsecured surfaces, but then you take even more care on where you're going. This robot doesn't seem to have the logic to do this, and that's where they're going wrong.

        What? That completely misses the point of this entire excercise.
        The point is to design and test a system that is capable of walking across such obstacles, just like you are, when it does become necessary. Once such a system is implemented, it will still make just as much sense to design the robot to avoid obstacles when it can, except now it can still keep going when it can't.

        In other words, these guys are trying to design a better support and locomotion system for future robots, instead of making a robot t

    • No human has vibrating feet.

      Then again, no human is born with metal feet with no sense of touch

      • by Anonymous Coward

        Then again, no human is born with metal feet with no sense of touch

        Yes, I always wonder if this isn't the root cause of failure to get robots to walk well. Granted, I'm sure someone's about to "OMG why would you think that experts wouldn't think of stuff like that" me on this, but I note that when I'm standing, I seem to mostly rely on my sense of pressure from my feet. I want to keep my center of mass over the center of my feet, so when the front of my feet feel more pressure than the back, I push the front of my feet down to cause me to lean further back, and of course

    • Re:The robot.. (Score:4, Insightful)

      by catchblue22 (1004569) on Sunday November 24, 2013 @01:10PM (#45507507) Homepage

      No human has vibrating feet.

      I beg to differ. Try this experiment. Place a thick pillow on the ground, or perhaps two. Stand on these pillows with one leg and no other support for one minute. If your pillow stance is unstable enough, your foot will have to move around rapidly to maintain your balance, since by not being able to change the location of your foot on the pillow, you must instead change the orientation of your foot. I suspect that this robot is actually testing specifically the ability of ankle joints to maintain balance, since there is almost no side stepping visible on the part of the robot. The Boston Dynamics robots referred to in other posts often seem to rely mainly on sidestepping for balance, and often have peg legs instead of feet and ankles. This robot seems likely to be a proof of concept of one particular method of balance that in future robots will be combined with other methods of balance.

  • by NixieBunny (859050) on Saturday November 23, 2013 @08:27PM (#45504285) Homepage
    Those flat board feet are no match for the terrain. A few metatarsals would go a long ways toward stabilizing its stance on uneven surfaces.
  • by Attila the Bun (952109) on Saturday November 23, 2013 @08:41PM (#45504339)
    An impressive balancing act, but the really interesting thing is to see how Atlas stumbles on objects which shift under its weight. It vividy shows the need for flexible feet on a biped. That's going to be an interesting engineering puzzle to solve.
  • by Okian Warrior (537106) on Saturday November 23, 2013 @08:47PM (#45504363) Homepage Journal

    The most recent Big Dog [youtube.com] video shows that this is a (largely) solved issue for quadrupeds. The middle of the video shows it walking over stumps, navigating a swingset, and so on.

    Still, "four legs good, two legs better". Or so they say...

  • Impressive (Score:4, Funny)

    by Brucelet (1857158) on Saturday November 23, 2013 @08:54PM (#45504397)
    The summary is an impressive bit of (unattributed) quoting from the article, evoking, the deepest fears of comma abuse.
  • by Anonymous Coward on Saturday November 23, 2013 @08:56PM (#45504405)

    hardly...
    It might freak the cat out in a similar vein to a vaccuum cleaner but wathcing the video makes me feel worried *for* (not of) Atlas

    I'm sure it's an amazing technical achievement but just highlights how much further we have yet to go.

  • by Anonymous Coward on Saturday November 23, 2013 @09:13PM (#45504473)

    Shouldn't the random obstacles include human skulls?

  • Difficult for a human to walk over those objects. A human would look down at the ground, and position their feet between the obstacles, or just walk around them. My wife would probably clear the stuff off the floor.

      I'm sure if I attemped to blindly walk across the bits of wood, rope and pebbles, I would probably trip over too.

  • When Atlas attacks, just throw down some piles of rope and cut 2x4s.
  • hell it made it father then i would have over that crap...bravo Darpabot!

    I, for one, welc...oh shit that's right.

    nevermind.

  • but how well does it handle a pile of human skulls?

  • Really? Even Honda's robot walks better than that...

    But they are all missing the point. The flat feet need to go. If you want to build a robot that walks like a human, give it human feet, give it an arch, give it toes, those are all important to how we walk and balance.

    Give it a body structure like ours, and it has a chance to walk like us. Why fight mother nature?

  • I mean, it's obvious that they're REALLY BAD at it. Hardly any animals are bipeds, in comparison to the amount of quadrupeds, especially.

    Why not play to robots' areas of strength and just stick with wheels/treads/rotors/wings/etc., and worry about more advanced forms of locomotion later? I understand that humans want robots that look and act like humans, I suppose, but how about coming up with more practical designs?

    • by fractoid (1076465)
      We can already do those things. They're solved. Hence, now we are at the "later" stage where we work on advanced forms of locomotion.

      There are several advantages to a humanoid robot in terms of the robot being able to function in a human-centric world (drive vehicles, use tools, navigate buildings etc.)
    • by tlhIngan (30335)

      Why not play to robots' areas of strength and just stick with wheels/treads/rotors/wings/etc., and worry about more advanced forms of locomotion later? I understand that humans want robots that look and act like humans, I suppose, but how about coming up with more practical designs?

      Because the world is built around bipedalism (because humans built cities, buildings, etc). If you want to navigate a modern city, two feet generally will get you everywhere, while the other forms will get you limited access at b

  • *Shrug*
  • Notice that every pace was exactly the same. There was no adjustment to place the foot in a better position. Just walk a pre-set pace and hope it can balance on the random spot the foot comes down. People do not walk that way and we fall down a lot less. I would also like to see that robot try to stand if it fell on it' back.

  • by Animats (122034) on Saturday November 23, 2013 @11:31PM (#45504981) Homepage

    That's the amateur version from the Florida Institute For Human and Machine Cognition. Here's the pro version [youtube.com] from Boston Dynamics, showing some walking over rocks. The balance control is better, the walking is faster, and the arms and torso are being used more effectively for balance.

    The "DARPA Humanoid Challenge" teams are struggling along. They had to write their software to run in a poor simulator, then use it on the real robot, with a competition next month in December. So the control software is crude. Most of the team efforts seem to be going into the perception side. Performance in the simulated humanoid challenge was poor; the best team fell down about 12 times. This looks like they're still using the basic balance controller from Boston Dynamics for control. Entrants in the competition get a closed-source .so file that will operate the Atlas robot for a few basic functions (slow walk, stand, etc.) for debugging purposes. This isn't the good stuff; Boston Dynamics keeps the better algorithms a secret. Entrants are supposed to replace those algorithms with better ones, but since they've only had about two months with the real robot, that probably hasn't happened.

    In a year, this will probably suck a lot less.

    • by Anonymous Coward

      You're fairly well informed, but not quite right. This is IHMC's own walking algorithm, not the Boston Dynamics stuff, as can be seen more clearly from e.g.
      this video [youtube.com] [youtube.com]

      You're right about the two months and about sucking less in a year though ;-)

  • To get to the other side?

Never put off till run-time what you can do at compile-time. -- D. Gries

Working...