Slashdot Mirror
Robot Unravels the Mystery of Walking
manchineel writes with a link to a BBC article on the lessons learned from a project in locomotive robotics. 'Runbot', as it is known, is the result of a modern technology combined with a 1930s physiology study into human locomotion. The study found that walking is largely an automatic process; we only engage our brains when we have to navigate around an obstacle or deal with rough terrain. "The basic walking steps of Runbot, which has been built by scientists co-operating across Europe, are controlled by reflex information received by peripheral sensors on the joints and feet of the robot, as well as an accelerometer which monitors the pitch of the machine. These sensors pass data on to local neural loops - the equivalent of local circuits - which analyse the information and make adjustments to the gait of the robot in real time."
Don't we need a crawlbot before a runbot, or did I miss something here?
libertarian: (n) socially liberal, financially conservative; neither left, nor right.
If there's something the world probably didn't need, it's another planar walker. Of course, the researchers are probably quite honest about the limitations when applying this to full 3d walking, but all that is lost in the translation to an article and then a slashdot blurb.
So why can't some people walk and chew gum at the same time?
Life is pain. Anyone who says differently is selling something.
Now, if someone could just describe the finger-arm reflexes needed to make a first comment post and implement that in some kind of program or robot thingy...
Everytime I read another study about how scientists have tried to replicate something humans find easy, and only manage to produce something that performs the task awkwardly, stupidly or otherwise ineptly, I feel vaguely in awe of how amazing the human body is.
Especially considering we appear to be a result of dumb luck and retarded fish monkeys..
libertarian: (n) socially liberal, financially conservative; neither left, nor right.
I'm getting some mixed signals from this article:
"How does Runbot walk?"
"The basic walking steps of Runbot"
"When Runbot first encounters a slope these low level control circuits 'believe' they can continue to walk up the slope without having to change anything."
"Runbot walks in a very different way from robots like Asimo, star of the Honda TV adverts, said Prof Woergoetter."
"The first step in building Runbot was creating a biomechanical frame that could support passive walking patterns."
"So using the information from its local circuits Runbot can walk on flat surfaces at speeds of more than three leg lengths per second."
"Prof Woergoetter said Runbot was able to learn new walking patterns after only a few trials."
"Runbot is a small, biped robot which can move at speeds of more than three leg lengths per second, slightly slower than the fastest walking human."
And last but not least:
"Four other scientists - Poramate Manoonpong, Tao Geng, Tomas Kulvicius and Bernd Porr - are also involved in the project, which has been running for the last four years."
Sorry guys, but it really isn't living up to it's name.
"Anyone who [rips a CD] is probably engaging in copyright infringement." - David O. Carson
"..controlled by reflex information received by peripheral sensors ..., as well as an accelerometer which monitors the pitch of the machine. These sensors pass data on to..."
S E G W A Y
Someone had to say it.
Hasn't this been known for quite a while? The actual task of walking is something that takes WAY too much computation (for lack of a better term) than the conscious brain is capable of. The same goes for quite a few other tasks that we perform. Think about image recognition, or throwing and catching a ball, or TYPING! Howabout READING!!
imagine a beowulf cluster of human brains!
NewslilySocial News. No lolcats allowed.
or does anyone else find it a little funny how they did a study on a task that's as basic as putting right in front of left and vice versa?
We came,we saw, we kicked it's ass!
The researcher's page on the robot http://www.cn.stir.ac.uk/~tgeng/research.html. Check the videos they are quite amazing.
I remember from my animal physiology classes seeing experiments about how cats walk. Apparently quite a few of the nerves which control the muscles used for walking can be severed prior to the dorsal root ganglion, and when placed on a treadmill the cats will still walk just fine even though there is no signal going from the brain to the muscles themselves.
09 F9 11 02 9D 74 E3 5B D8 41 56 C5 63 56 88 C0 is the magic number.
The British have been working on this for years!
http://www.youtube.com/watch?v=IqhlQfXUk7w
It's nice to see the Runbot "has been built by scientists co-operating across Europe".
Operator, give me the number for 911!
robot did not unravel the mystery. it merely reinforced the idea that local processing can make robots walk in an efficient manner. mystery was unraveled 70 years ago by Russian physiologist Nikolai Bernstein.
Ants have a fully autonomous walking sub-system. Here is how you find out:
1. Arm yourself with a box cutter, straight razor, razor blade or scalpel
2. Capture your favorite back yard ant.
3. Cut off the ant's head. Be careful not to hurt anything else, don't smash any legs and don't crush any other body parts. If you don't get it right with the first try, try again on your next favorite ant.
4. Discard the head as neither you nor the ant can use it anymore.
5. Let go of the rest of the ant
The ant should now right itself and stand as if awaiting movement instructions.
Some fun experiments:
1. Blow gently on the ant. It should sway in the breeze but generally remain upright.
2. Flick (or blow harder on) the ant without smashing it so that it tumbles some distance. It will right itself and patiently await further instructions.
3. Place the ant on a piece of paper, wait for it to right itself and then flip the paper over. The ant should stay attached to the paper.
Ants are truly miniature engineering marvels.
3 leg lengths per second is just short of the speed of the "fastest walking human"?? Somehow I doubt that. Racewalking is an Olympic even, even, and I know that some folks can do like a 6-min mile walking. Assuming a leg length is a yard, that robot would take closer to 10 minutes to walk a mile. So... it's kind of a dubious claim.
I like basketball!!1!
What about sex?
When will they automate that?
Great research! Where can I buy a bipedal robot kit with this technology for my next robot project? Oh, I can't. Too bad.
So if I want a bipedal robot I have to duplicate your work. Maybe I can read your scientific papers and that will give me 10% of the knowledge you gained in doing this project, but I still have to turn theory into practice.
Commercialize your research already.
How we know is more important than what we know.
You are missing a critical pice in your statment.
it should be:
"Especially considering we appear to be a result of dumb luck, retarded fish monkeys, and time.."
People just can't or don't take time into account naturally. You see it all the time.
The Kruger Dunning explains most post on
This weekend (with my bottle of tequila) I'll be testing the mystery of falling over.
http://dx.doi.org/10.1371/journal.pcbi.0030134
Science has proven that health-weenies in the form of joggers are mindless.
The feedback system is certainly a step in the right direction as well as using the idea of the "falling forward" concept of walking. For the other posters who stated it "walked funny" you didn't notice that the ankles of the Runbot are locked in position and it merely rolls on its "foot" until the next step. You try walking with your ankles in a fixed position and we will see if you don't look funny doing so. If the feedback system can be extended to small motors in the ankles I think the appearance of the Runbot walking would be a lot more realistic, and therefor a lot more acceptable to the anal, obcessive-compulsive Slashdot community.
Why the hell don't they tell us this kind of stuff sooner? I've been forcing myself to think, "Right foot, left foot, right foot, left foot... for going on 25 years now!"
It is no longer uncommon to be uncommon.
I think one of the main problems so far with walking robots is that while they can move their joints and things accurately, they can't fine-tune the movement very well. I think the first step might be to add sensors on the feet. It might seem strange, but us humans can feel how much weight is on either leg. Until the robot can detect how much weight is being placed on each foot, I doubt it'll be able to walk with the proficiency of a human.
This seems to me to be of philosophical interest. Namely to the frame problem, and objection to artificial intelligence theory which claims that a computer isn't capable of efficiently ignoring information that's not of immediate importance. That's exactly what this robot does, though. It should be interesting to see if any debate comes out of it.
I'm pretty sure we know how walking works and how humans came to walk the way they do- evolution by the process of natural selection. Teaching a computer to "evolve" in the same fashion is nothing new- they've been doing this stuff for decades. Heck, one of my AI class projects involved genetic programming where the most *fit* code was passed on to the necxt generation.
Any act you repeat frequently enough becomes partially hardwired into the nervous system, and we call it "muscle memory" (though of course it is neurons that retain the memories). If you have ever learned to play an instrument beyond the beginner level, you will know that you cannot possibly process everything that needs to be done, in real time, in the conscious mind. At some level, you have to just put it on autopilot. You need the conscious mind to read the chart or pick out the harmonies, but you expect that the skills necessary to translate your ideas into sound will just be there. If you're thinking "how do I play that note", it's already gone by.
If you want to play an instrument and sing at the same time, or play two independent instruments at once (piano and especially organ are close enough to qualify, as is something like a Chapman Stick or Megatar), you have to rely on muscle memory that much more, as you now have twice as much to deal with. Doing all that and singing at the same time is more difficult still, and there are plenty of great musicians who never learn this particular stunt. The only way I can play and sing at the same time is to drill one or the other (usually the instrument) until I can do it by habit alone, then layer the other one over it and hope it holds together. Fortunately, woodwind players are not frequently asked to sing while playing, or to play two instruments at once, and if I do have to sing while playing, it's not really an independent act but part of coaxing a particular sound from the instrument.
As is the case with walking, the trick is to practice (a lot) and to accept that you will fall down (a lot) until you get the hang of it. Most of us just don't remember how hard we had to work to learn to walk. Some have to re-learn and could tell you how tough it is, and others still bear the scars of learning in infancy -- I have a scar in one eyebrow from falling into the edge of a table while still learning to walk (and a matching one in the other eyebrow, from learning to fight, but that is another story).
Mal-2
How is the Riemann zeta function like Trump rallies? Both have an endless number of trivial zeros.
More natural-looking (albeit slower) performance from the Wabian-2.
Swiveling hips are the way of the future. ^_^ Here is a demonstration video. (The giant mech shooting balls at people afterward is unrelated...)
Also check out the related robot Kiyomori. Because nothing says "We are here to protect you" like traditional armor and GLOWING EYES.
Don't put advice in your sig.
The author of "How to Survive a Robot Uprising" says marching 'bots like RunBot won't be terrorizing our towns anytime soon. We sure about that?
was an automatic process.
"All these big machines stomp around like robots - we want our robot to walk like a human."
Based on what I've read and seen, this article is wrong about the Asimo. The Asimo is the only robot I've seen that looks very human in the way it moves. It can walk, run (with both feet leaving the ground), jump, perform a complex dance, get up after a fall, adapt to changes in the terrain, and maintain its balance if something unexpected pushes it. It also treats walking/running as a controlled fall.
It looks like runbot can't even get both feet off the ground, which means it's not running, it's power-walking. The only thing new here may be its "local circuits", which simply means that it has extra CPU's to take the load off the primary CPU.
The video was very grainy and didn't play for me. Here's the same one at YouTube.
Patrick Doyle
I mod down every jackass who puts his moderation policy in his sig. Oh, wait a sec....
It has long been known that much of walking is reflex-driven. A decerebrate cat (i.e. with the brain disconnected) supported on a treadmill will go through normal walking movements (known as "fictive locomotion") and will even correct for "stumbles". It seems likely that the timing required for coordination of walking is simply too tight for the brain, with its longer transmission delays, to manage properly.