Forgot your password?
typodupeerror
Science Technology

Universal Manipulator Does Chess 88

Posted by timothy
from the good-girls-from-bad dept.
SillyWilly writes: "A while ago a story was posted here about a vibrating plate capable of sorting color poker chips, and there was much ado about the videos being real. Well, a new cool demo involving chess pieces is out, done by a fellow called Neil Aldrin, who is hacking away at Dan Reznik's (the original inventor) cool contraption."
This discussion has been archived. No new comments can be posted.

Universal Manipulator Does Chess

Comments Filter:
  • by Anonymous Coward
    I used to do this with little rubber Kinniku men on a little vibrating wrestling platform with my friends... (that didn't sound good...)

    Before that, there was a little football game that did the same type of thing.

    This is cool?
  • by Anonymous Coward
    >Example? Look at Tribes 2. This game is true beauty. You won't ever see teams of bots capable of beating trained human teams.

    I agree. And you won't ever see a computer that can beat a human at chess. Oops, sorry, wrong decade

  • by Anonymous Coward
    As a kid, I always hated that damn football game. I'm a nerd, so of course real football is out of the question... but playing the stupid board game, my quarterback ALWAYS ran in the wrong direction. I almost think my brother found a clever way to cheat. But thanks to the wonders of technology, not only will he run in the right direction, he can also dodge and weave around the defensive player that the dog chewed the head off of.

    As long as my brother's pieces still move randomly, this is ultra cool.
  • by Anonymous Coward
    With all these posts about Bill Gates, don't we have enough articles dealing with manipulators? Let's see something more positive, please.
  • the pen is mighter then the sword. the sword is mighter then the court. the court is mighter then the pen.

    Ah, but the spell-checker is mightier than the sig file.
  • by Anonymous Coward on Tuesday July 31, 2001 @02:07PM (#2177998)
    Are you sure it wasn't Buzz Armstrong? :)
  • by clem (5683)
    I suppose a human would have to sit at the board, if only so the knight could hop over other pieces...
  • spinning is easier than moving them, in many ways.

    Lea
  • I'm actually not working for PARC right now (UC Berkeley instead -- crypto is fun!), but I don't see how putting that in my .sig would encourage other women in any case. Feel free to email me to discuss this. (put an informative subject so I don't delete it as part of the massive purges on that account)

    Lea
  • oh dear me no, it's far more complicated than that. I won't get into details, but basically you create a "force field" by vibrating the table in the right x, y, and \theta... this hasn't been worked on enough, as I see it.

    Lea
  • by chialea (8009) <chialea.gmail@com> on Tuesday July 31, 2001 @01:51PM (#2178003) Homepage
    The movies aren't faked. I used to work for the same professor (John Canny), and I saw the table in action. It's pretty cool stuff, though it could go a lot farther with some more math work on the field created.

    Yes, the table is real. Yes, it works. Yes, it could be better. This is research... this is expected!

    Lea (who used to work on cs theory and robotics, and now works on cs theory and crypto)
  • I'd like to see the magnet that can lift an aluminium can. Or a piece of glass, for that matter.

  • I'd like to see some better videos of the movement and maybe some information on the other equipment. The videos are of such high-compression that they look to be faked already (though it is apparent that lots of compression is being done)

    problem is I can't forsee any real practical application of this technology (item sorting is great, but why do it this way? sure, for HazMats it _might_ be useful, but looks extremely complicated.)
  • by BeBoxer (14448) on Tuesday July 31, 2001 @02:21PM (#2178006)
    but I can't do this. What's cool about this isn't the fact that they have figured out how to have a computer sort poker chips. It's the fact that they can do it by simply shaking the table! I challenge you to find any human that can simultaniously pass the salt to one end of the table while passing the pepper to the other end of the table by merely shaking it! There is nothing humanlike about this at all.
  • That is correct, Dr Falkien.

    Now I will not enact Global Thermonuclear War.
  • from the good-girls-from-bad dept.

    None of the links provided had to do with good girls, or bad girls. And I was ready to get excited, too.

    --
  • by Elwood P Dowd (16933) <judgmentalist@gmail.com> on Tuesday July 31, 2001 @08:24PM (#2178009) Journal
    If I'm not mistaken, it doesn't do any "piece recognition." The pieces are set in predefined layouts, and the computer just has to watch where they move. And the reason to be overwhelmed is assembly line production in places where robot arms and humans couldn't work.

    The question, though, is if they can spin the chess pieces. If they can do that, then they've got a considerably more useful tool.
    --
  • i would just like to thank you for the off-topic link to the kasparov deep blue blue story. seriously. it was a great read. i play horrible chess, but i enjoy watching and learning about it.

    complex
  • by jetpack (22743) on Tuesday July 31, 2001 @03:40PM (#2178011) Homepage
    Well, I used to work for a prof that did research in this area. That was five years ago, so I'm way out of date, but hopefully I can give you a bit of insight.

    A big use for this kind of thing is for part feeding, that is a method of orienting parts on an assembly line. There is mention [berkeley.edu] of this on Dr. Resnik's web page. Basically, you have the problem of presicely orienting a whole whack of (possibly complicated) parts as fast as you can to present them to the next stage in an assembly process.

    Yes, a robot can accomplish this, but because of the motion that is involved they are slow, and because of the optical recognition involved they can bit, uh ... touchy :)

    One method of dealing with part orientation is by things called bowl feeders. They are a bit hard to describe in type, but imagine that you have a big steel drum with a spiralling track up the inside of the drum. By vibrating the drum you can make the parts you are trying to orient climb up the track (beleive me, this sounds wierd, but it actually works). By changing the shape of the track you can force incorrectly oriented parts to fall back into the bowl, thus filtering out parts that are correctly oriented. So, only parts that are correctly oriented arrive at the top of the drum.

    It's not quite as simple as that, but that's the general idea. Now, as well as this works (when it works), the problem is that whenever you change the shape of the part, you need to build a new bowl feeder! And building these things is not simple (or cheap).

    I beleive what Dr. Reznik is trying to do with this experiment is not to prove that you can move poker chips around, but that you can build a programable solution to this problem; you can build one machine that will sort anything, given the correct programming of the controllers, thus alleviating the cost of prototyping things like bowl feeders.

    So, to my mind, it's actually pretty spiffy, despite what others are saying here.

    I know I've left out some details, and I certainly haven't researched Dr. Reznik's work, but hopefully I've given you (and some of the /. detractors) some idea where this work is probably headed.

    The only thing I find a bit perplexing is, if he is proceeding towards the end that I've described, is how he is going to do this in three dimensions. Sure, he can rotate things in two dimensions, but what about more complicated parts?

    I should also add that watching one of those bowl feeders in action is actually pretty creepy at first. Parts move up the track, but nothing else seems to be moving (the vibration rate is quite high, so you don't really see it). Mind you, they are freakin loud :)

  • IANAP, nor have I read the thesis but I would guess the rotational vibration can be made to interfere with the translational vibration in such a way to produce a net positive motion (see parandos paradox). Varying the phase difference would allow control over the induced motion. It doesn't work for other objects because away from the center of rotation, the rotational vibration "looks" like a translational vibration. Just a guess...
  • Same here! why isn't this one modded up to funny????

    Maybe the moderators were having a bad day...
  • Mathematics!?

    I'm suprised. I has just assumed that they were using some dynamic feedback system, so that when the peice moves in the right direction, they keep it up, and if it moves in the wrong direction, they do something else.

    I'd expect that a neural net driving a set of vibration generators would work quite well.

    hrm?
  • The basic idea is that a rotational vibration centered on the point at which motion is desired is superimposed on a translational vibration in the desired direction of travel. When both vibration functions are suitably chosen, there's a very unexpected property: the feeding velocity is small everywhere except near the center of rotation.

    This is counterintutive. You'd expect the rotational effects to be biggest far from the center of rotation, and zero at the center of rotation. Apparently the idea is that the forces induced by rotation interfere with the translational vibration that makes objects move. What puzzles me is that they're able to achieve zero feeding motion over most of the entire plane. But look at figure 6.2 in the thesis, showing the jet field.


    Yes, objects far from the center of the jet still see a small force. But you can also apply a "jet" to each of them to cancel the small side-wash from the main jet. Or just wait until the side-wash displaces them from their desired locations enough to matter and then "blast" them momentarily with their own jets to put them back. B-)

    That's really neat. But I don't get it intutively yet. Can anybody else explain it more clearly?

    It's brilliant! (He really deserved that PhD.)

    I think I get it:

    The key is mentioned elsewhere in the paper: Once you've broken the static friction the force is constant (the sliding friction) in the direction of motion. So if the motion of the underlying platform lasts longer in one direction than the other, you have a net force in the direction that the motion lasts longer. (You move the plate faster in the other direction to compensate, so it ends up back where it started. The object sees a force vector which is the vector integration of a a vector that is pointed in the direction of the platform's motion but "clipped" to a constant length.)

    That's the description of the linear part of the motion. But there's an additional rotational motion. The object at the center of rotation doesn't "see" it. But as you get farther from the center of rotation, this component becomes progressively larger. The motion of the table is the vector sum of the two motions. The force seen by an object is in the same direction, but "clipped" to a constant magnitude (vector length).

    As you get farther from the center of rotation the rotational component of the table motion increases but the translational component remains constant. So the rotational component provides progressively more of the vector's direction. But the force is sliding friction so its magnitude is limited to a fixed value. So for the force vector the rotational component quickly becomes the bulk of the force vector, causing the translational component to decrease. The rotational component becomes dominant at a very short distance from the center.

    But though the translational component is asymmetric, conveying a net force over one cycle, the rotational component is symmetric, conveying no net force. So you're presented with a net translational force that falls off quickly with the distance from the center of rotation.

    CUTE!

    All of this neglects another factor: At some points the table will, for some part of its cycle, match the motion of the object over that point. The object will "grab" the table and get an additional shove from the static friction as the table accellerates.

    The demo seems to show a device that neglects this effect initially, then compensates for it by applying additional "jet" force on any object that is displaced from its desired trajectory, returning it to its desired location. The result is some "noise" in the motion of the individual objects and feedback correction of it.

    It seems to me that this is something that can be turned to an advantage, increasing the effectiveness of the jets. If the object near the center of rotation is allowed to grip the table it can be accellerated more effectively than if it is kept sliding for the full cycle. Meanwhile, objects far from the center are kept in sliding-friction mode. The result is again a "jet", with the at-center object experiencing a shove and the off-center objects having the shove killed off by the rotation. But the shove might be stronger and the cancelation more complete.
  • The applied force is clipped by the limits of static friction.

    Sliding, not static. Otherwise, yes.
  • The key to the jet: rotation is symmetric [a pure sine wave, sin(2t/3)]; the translational vibration is not [cos(t)-cos(2t)/2].

    Nice. That should keep everything sliding continuously, or very close to it. Uses three frequencies (2, 3, 6).

    Do you get multiple jets by doing linear addition of single jets at different frequencies or do you switch between single jets in sequence? If the former, is there a nice rule for the frequencies to prevent nasty effects (such as unplanned motion from static friction)?

    (Please excuse my laziness if you had that in the paper.)
  • What exactly is spectacular news here? i mean pepole make all sorts of fantastic programs/hardware for their ph.d, and this is what slashdot features?

    Im not trying to rain on anyones parade or anything, but this is a non-event.
  • by Ukab the Great (87152) on Tuesday July 31, 2001 @02:57PM (#2178019)
    If I could manipulate entire universes, I'd find chess kind of boring.
  • by Greyfox (87712) on Tuesday July 31, 2001 @04:00PM (#2178020) Homepage Journal
    I find most of the applications in the original article a bit forced. Self setting tables (Assuming you do 95% of the work and move the dishes to the table and chuck them down anywhere.) Conference room tables that serve coffe (Again, assuming you do 95% of the work to get the coffee to the table.) The coolest thing about the chess thing is the piece recognition the computer does, which a camera is used for. The Post Office application is, admittedly a bit cooler but for the most part I'm not really overwhelmed. Just kinda whelmed...
  • People are correct in saying that robotic arms would be able to move bigger objects and is probably a cheaper solution. Apparently, UPM allows movement of more than one object at a time in a small area. If you have a robot arm, let's face it: you need room for it to move. This may be room that you don't have.

    Having said all that, I still don't know exactly what real world applications this may have. Would it be used to sort drugs? Maybe someone can think of a good reason to move multiple pieces of something at the same time. This is somewhat exciting, but we're years away from any practical application, especially since I can't even begin to think of one. :)

  • A watch factory is an augmentation of a watchmaker. Someone programmed the robots in the factory, after all. If it's possible for a small group of people to make many watches by building a factory, should a society really waste perfectly good people on the task of moving small gears into specific arrangements? Much better that those people should be able to get on with tasks that aren't so, well, mechanical. If a persons job could be eliminated with a machine, how rewarding could the job have been? Now, if you want to argue that we don't allocate sufficent resources to education, so that a replaced worker can't go back to school and learn something else more interesting, I'll agree with you there. But there's too much fundamental work to do in the world -- getting food to hungry people strikes me as a good example -- for there to be anything wrong with the elimination of jobs. The problem isn't a shortage of work, it's a failure of organization.
  • Being at UC-Berzerkley :), does he have to relinquish his thesis to UMI for publishing? That would be a shame.
  • Yeah, but if it's a sparkling beverage (lager, champagne, Coca-Cola) it'll fizz right out of the glass!
  • It's a machine that sorts. Go figure :)
    --
  • hot damn, i was thinking the same exact thing!
  • But the force is sliding friction so its magnitude is limited to a fixed value. So for the force vector the rotational component quickly becomes the bulk of the force vector, causing the translational component to decrease. The rotational component becomes dominant at a very short distance from the center.

    Ah! That's it. The applied force is clipped by the limits of static friction. That's the nonlinearity that makes it possible. Very neat.

    I agree that this guy earned his PhD.

    This will have industrial applications in manufacturing and material handling. It's a simple, robust device to build.

  • by Animats (122034) on Tuesday July 31, 2001 @03:16PM (#2178028) Homepage
    That's very impressive. I'd have thought that impossible.

    Recognize what they're doing. They're vibrating a rigid plate in such a way that one object, out of many, moves. The system that drives the plate can produce both small rotations and translations. The vibrations have arbitrary waveform, and are generally asymmetric. It's the asymmetry that produces motion. That's all.

    I could see this working for two objects, because you could vibrate the plate such that the center of rotation was under the one you didn't want to move, so it didn't go anywhere. But I had no idea how they make this work for N objects.

    The novel result in the thesis [berkeley.edu] is section 6.1. Figure out how a "jet" vibration works, and you'll understand the whole thing. The basic idea is that a rotational vibration centered on the point at which motion is desired is superimposed on a translational vibration in the desired direction of travel. When both vibration functions are suitably chosen, there's a very unexpected property: the feeding velocity is small everywhere except near the center of rotation.

    This is counterintutive. You'd expect the rotational effects to be biggest far from the center of rotation, and zero at the center of rotation. Apparently the idea is that the forces induced by rotation interfere with the translational vibration that makes objects move. What puzzles me is that they're able to achieve zero feeding motion over most of the entire plane. But look at figure 6.2 in the thesis, showing the jet field.

    That's really neat. But I don't get it intutively yet. Can anybody else explain it more clearly?

  • Its posts like this that make me wish I currently had moderator points ;)

    -= rei =-
  • Cool - How about putting some magnets in each piece and then disabling the magnet as the piece is about to move. That way only the moving piece is, well, moving?

    --
  • Watch the chips when they get right to the edge *poof* they vanish. Not a single chip falls and lands outside the table border. I watched both the realvideo and mpeg versions. Someone knows adobe premiere well.

  • Wow I'm honored that my comment pissed off some moderators enough to get marked as flamebait. Oh wait don't tell me there are quadriplegic moderators out there.

    -Nails-
  • I've met the guy. The name is not a random coincidence =)
  • Damn you, that's what I was going to say!

    At least someone else picked up on his name.

    Hacker: A criminal who breaks into computer systems
  • Look at the page a little more carefully before you complain.

    ---
  • Thanx for teh corection! ;)

    Seriously, I hadn't noticed. I appreciate it.

    ---
  • Yeesh. Normally I wouldn't bother posting this, but I've read 4 comments about how the propriatary bastard is offering the videos only in windows media format! Good lord. Read a little further. It's in good ol' MPG1 too.

    They clearly only are posting in windows media because it uses so much less bandwidth. Actually windows media uses MPG4 as it's underlying implementation, which you can play in linux anyway if you really care to. Yeesh!

    ---
  • by coupland (160334) <dchase&hotmail,com> on Tuesday July 31, 2001 @02:15PM (#2178038) Journal
    I hope someone who understands robotics better than me can explain -- what is so special about this device? I saw the chess and poker chip demos and I was impressed, but the same could be accomplished using a robotic arm, probably in less time. Can someone in the know explain the import of this technology? I'm sure it exists but I haven't read enough on the topic to see it plainly...
    ---
  • http://www.cs.berkeley.edu/~dreznik/UPM2000/images /sorting/sort1.ram [berkeley.edu]

    You'll notice that the chips do fall off immediately, but in the last few seconds (and the first few) of the clip you will see something in the top right of the view come in the shot very quickly then leave. This leads me to think that the shots were sped up intentionally to not bore the /.'ers, and keep your attenetion.

  • Manipulating N objects simultaneously isn't possible. As it (kind of) says in one of the articles above, the system only moves one object at a time, but only a small amount. Then it moves to the next one, and so on, just really, really fast so it seems like all are moving at the same time.

    This is similar to how your monitor works. It doesn't turn all the pixels on at the same time, but one at a time, it just switches really fast. The result is what seems like a continuous image.

  • Whats so special about this device? WHATS so special about THIS DEVICE??

    My God man.. haven't you ever wished that the vibrating Football player game actually worked?

    "Put a paperclip in the QB to make him run" my ass.. :P
  • I have seen something like this at work sifting grains into piles of type. This idea looks cool and it's applications look immense. When I have seen them sorting alu cans from glass, it is usually done with a magnet, the complexities of sorting dusts and grains is much more impressive a feat IMHO.
  • ...am I the only one reminded of those cheesy "magic fingers" hotel beds?

    Think of the new applications.

    (No, not those, get your mind out of the gutter, get married for a few years and rejoin us at the grownups table.)

    The bed automatically gives you back your half of the blanket when your spouse (significant other) steals it.

    The bed automatically shakes cookie, cracker, and granola crumbs out of the bed after your toddler finishes watching cartoons on your TV.

    The bed automatically finds the remote and returns it to the head of the bed.

    Um, well, OK, so the bed would also have ... other uses... which would probably add to the granola and cookie crumbs in your bed.

  • Kinda reminds me of the old electric football game where the pieces just vibrated all over the field. Of course, the damn thing never worked right... :) Maybe after all this time, some game manufacturer can make an electric football game that actually works!
  • This machine knows how to sort chess pieces.
  • The "specialness" is that it manipulates objects specifically *without* using a "grabber". It basically buzzes or shakes objects from one place to another across a surface. The amplitude and direction of vibrations can be varied, which allows for some really cool effects like moving only items that weigh within a certain range, or are of a certain shape. Example. Pour some salt, some flour, some ball bearings and some little plastic fishing floaters into a bottle. Shake well. Tip the homogenous mass onto the manipulator surface. It can be programmed to send different vibrations in different directions which will have the effect of sorting the four different kinds of stuff into the four corners of the table. Very very cool. As with all new techniques, applications already exist and more will be found. Darryl.
  • Yeah, it would have helped if they'd at least superimposed the image of the board on the video. All I could see is a bunch of things moving randomly on a table, some occasionally falling off. Give me a camera and an earthquake and I can do that, easy.

    The pennies moving in the figure 8 is kinda neat, tho.

  • Hmm. I wonder if those penis mightier's really work.
  • by GiorgioG (225675) on Tuesday July 31, 2001 @05:15PM (#2178050) Homepage
    Now I don't have to get up if I forget my beer on the other end of the coffee table
  • The important fact about this device is not that it moves poker chips around, or that it can be used to set tables (what ? anyway), but that it is programmable. Meaning that it can potentially solve any problem which involves moving multiple objects to some specific configuration. What do you think is easier - building a physical robot arm capable of moving 100 objects at the same time, or writing a computer program using pre-built library functions ?
  • what you would use it for - to fill in the "further funding is required" checkbox on a grant....
  • when computers and machines do humanlike tasks such as basic sorting of real world objects... and they don't get all hot and bothered when they do non-machine like tasks, such as lifting big cars and things?
    Two reasons
    • Because "non-machine" like tasks has been done for a large number of years (it's not new)
      • From a technical (and object oriented :) point of view, once you have the machine to lift the big car, the code would look something like very_big_thing.lift(car); However, it is very difficult to actually recognise objects with a computer if you hook it up with a camera...

  • I wonder if they have patented this yet. After all, it would seem, they stand to make a lot of money by making possible commercial applications.

    Of course, the patent will probably be owned by the university, but universities license these kinds of patents out all the time.

    Just wondering...
  • I'm sorry, but I think the video clearly showed how un-prepared this UMP was... I saw two pieces move at the same time, several times...

    How about putting Chess pieces on a mattress while people get their grove on?

  • when computers and machines do humanlike tasks such as basic sorting of real world objects...

    and they don't get all hot and bothered when they do non-machine like tasks, such as lifting big cars and things?

    Robots that walk like people, human facial expressions on computer-graphics simulation.. they all generate so much interest among technical people.

    Perhaps, not being a technical person, I am more interested in the great alienness of machines. I am interested in earth movers at mines; the Big Dig in Boston; construction cranes; auto assembly lines; mainframes; enterprise-class servers; billion dollar electronic fund transfers.

    The replacement of humans in industrial processes with machines has always been an object of industrial design. The assembly line replaced the guild style craft. Instead of creating little portable machines that aided the watchmaker and the bootmaker in their old craft methodologies, inventors remade the ways of creating goods; those guild methods were replaced by roboticized methods of manufacture.

    I would rather see a chip in the head of the watch maker that gives him eagle eye vision, or an augmentation of intelligence or emotional sales skills in a salesperson, than something that would eliminate these people from their industries.
  • Yeah, and then you can put a bluetooth inside eveypiece so they can create a WAP network and communicate with eachother on ambushing the other team, and then we can put a bluetooth jammer on the queen's head so it'll mess up the communication between the other team and then we can give the little chess pieces guns, and then we can make the chess board have elevations and then we can put obstacles in the middle of the board so the pieces can hide, and then we can give the pieces little optical sensors so they can detect team members (wont be very hard, white or black) and obstacles and make a 3d model of the chess board using stereo vision, and then we can make the chess pieces walk...

    Or we can just play chess using plastic pieces on a metallic board and use our time making more useful things. But that wouldn't be cool would it?
  • ...the vibrating hotel bed industry.
  • That wasn't the point of the demonstration. We already have computers that can play chess. This was to demonstrate the movement of physical objects though vibration. They could have used self-dealing cards or a self-setting dinner table. At least try staying OT when trashing an article.
  • I had an electric football game back in the mid 70's that did pretty much the same thing. You set up the pieces, plugged it in. It buzzed and shaked, and the little guys moved around. A search for "ELECTRIC FOOTBALL GAME" on eBay turns up lots of the things.
  • Chess, as a game, is overtly mathematical. Chess as an art is boundless and never duplicated.

    I think that experiments like this forget the latter. The idea that a computer can duplicate moves programmed into it isn't all that interesting (aside from the complexities of the actual programming). This is not, as never should it be set out to be, a replacement to people or a way to show off machines and programming.

    I'm still puzzled as to why people feel the need to use Chess in these sort of matters. The great ones - Morphy, Nimzovitch, Spaskey, Fischer, Kasparov, Karpov etc - played the game as passionately as they did logically, and that's what made them great.

    Passion can not be duplicated or programmed into a machine.

  • Can't everything you mentioned be placed into a "sorting" catyegory?

  • by phantumstranger (310589) on Tuesday July 31, 2001 @01:47PM (#2178063) Homepage
    Deep Blue + UPM = an even more frustrated [cs.vu.nl] Kasparov? Do you think he would give the finger as he stormed out of the building? Do you think the machine would give one back?

    I'd watch just to find out.

  • a game of chess that plays itself. Doesn't that defeat the purpose of playing chess? Next up, self watching TV...
  • I disagree. A team of properly programmed bots, complete with 100% accuracy using hitscan weapons and almost unhittable with projectile weapons would be unbeatable by a human team. You don't need strategy to win if you never miss and can't be shot.
  • That's a pretty cool idea but I think it would be better with the magnetic squares on the table that could be activated and deactivated, that way you wouldn't have to change the design of whatever the product was that was being sorted.
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  • Don't you think it's time we stop slashdotting these researcher's websites? Please mirror these things before you go posting links like crazy. These people don't have enough time or money as it is, I don't think we should add to their headache by crashing their servers.
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  • It's at high speed. Watch the one at regular speed.
  • by danyelf (449491) on Tuesday July 31, 2001 @03:32PM (#2178069)
    Ok, first the full disclosure. I'm a grad student who worked with both John Canny (the advisor involved) and Dan Resnik (the builder of this cool device). http://www.cs.berkeley.edu/~jfc/

    The objections that I'm seeing seem to be of the following:
    * "It's not real--the videos are fake."
    No, it's real. The basic principle is this: if you take a table and shake it, stuff moves a little. If you shake it in a different direction, stuff moves in a different direction. Dan has figured out a pretty cool way of quantifying all this. He does a vector addition of several different shakes, and is able to therefore target the shaking.

    * Why is it cool?
    Well, first, because it's not obvious that it works. This is tricky math.

    * Yes, but why THIS?
    Because robot arms are a pain, and only manipulate one thing at a time, and they need a lot of elbow room, and a lot of motors, and they have to touch things. This requires four motors that JUST pulse in and out. It works on a flat table. Nothing touches except the table surface. This would be perfect for carefully-controlled environments like clean rooms (where you want to minimize the amount of stuff in the room), hazardous materials, and delicate objects. There's very little complexity, and because it's just a bunch of (tuned) vibrations, you could slap up a new one against a floor and it works.

    The chess demo is just showing that one can comfortably manipulate a large number of items.

    * Does anyone need to sort multiple things?
    All the time. Factory floors separate out rejects from working models. Recycling centers separate cans from bottles. Usually, they hire people to sort the stinking messes apart, and they use clever special-built machines to separate metal (use a magnet) from glass (heavier than plastic when crushed) from paper. This sorts on a smooth--therefore easily cleanable--surface.

    Dan now works for Siemens TTB, who are, among other things, very interested in small motors.

    * My toy did this.
    Yes, it did. But it did it in an extremely constrained way, and it probably took a really practiced flip of your wrist. And it probably did it in one dimension. (Dan has a little plastic train set that works on this principle).

    * This eliminates human jobs
    Not necessarily. It could work well in conjunction with a human job. Why do that annoying RSI-inducing reach/grab/sort when you can sit behind a desk, look at a video camera, and tick off the items on a screen? After all, image recognition isn't too good yet. The machine is responsible for the reach/grab/sort, and you don't have to wear a bunny suit.
  • by MarkusQ (450076) on Tuesday July 31, 2001 @06:55PM (#2178070) Journal
    Call me a skeptic, but I would have been much more impressed if they had moved one thing at a time. Instead, if you watch the video, you see them moving everything on the table at once, with no real pauses between steps, and most of the pieces in multiple squares at once. There is also considerable collective drift. The text speaks of accurate positioning of individual pieces, but that isn't what I saw.

    -- MarkusQ

  • there is also mpg if you open your eyes and look

    I have a little machine that seperates my quarters, dimes, nickles and pennies which is much cooler than this plate.
  • The way to think about this intuitively is to think about your "knowns" in the equation, and how to solve for your "unknown". Your "knowns" are the following: 1)You have chess pieces set at initial positions, 2)You know where you want these chess pieces to go (final positions), and 3)you can get info on the materials involved as they can act as "dampers", and dampers affect the vibrations. Your "unknown" is the forced vibration input (i.e. how fast each motor spins at a given time). For those of you with a calculus background, you can see that this is a second order linear equation with constant coefficients ( g(t)=ay''+by'+cy ). Thus you can set up matrices using Mathcad, etc. to crunch the math...solving for the motor inputs for each unit of time.
  • yet another machine that will be able to beat me at chess.
  • Haha kinniku men. Awesome. Although of anyone on Slashdot remembers them, they probably remember them as M.U.S.C.L.E. men.
    I wish I still had those. I don't know what I'd do with them but hey at least I'd know where they were.



    ------
  • but I bet it would be really good at pinball :D




    ------
  • Hi everyone, it's cool to see that my work has received so much attention. Thanks to Robert Cicconetti, I now have a video with an overlaid chess-board, so you can see where the pieces are in the chess game. I also changed the video links to list the mpeg's before the windows media files since many people didn't notice that I had mpegs. The reason I have such high compression is I have very limited web-space. --Neil
  • The key non-obvious feature of this device is that it has only 3 degrees of moving freedom [vibrate along X, Y, and rotate] and yet it can manipulate [displace, rotate] N objects, where N is arbitrarily larger than 3. One interesting use for the technology is allowing manipulation / parts feeding without overhead, pick-and-place like actuators, resulting in cost savings and added functionality. One other interesting use of this technology is for novel haptic interfaces. For example, a beer counter that slides a beer glass to a customer's hand upon a gesture. This demo is being planned for the near future. -Dan Reznik
  • You can't solve it this way because the table only has 3 degrees of freedom, and you are imposing 2N constraints, where N is the number of parts you're translating. That is the beauty of the method proposed, i.e., a deceptively simple device achieves great dexterity. To solve this problem, you have to think in terms of the average sliding friction field resulting from a complete closed horizontal vibration.
  • Actually this is a subject of further research. I have looked into using more sinusoidal components in the X,Y,theta signals to achieve non-zero force focii at two or more spots on the table simultaneously. --Dan Reznik
  • What a disappointing reaction from /.

    This is real world engineering. It may be trivial to write software to manipulate objects on a GUI, but this application requires indirect mapping between vibrating transducers and deterministic motion of 1 object whilst N objects remain relatively stationary. Iterate this and you have arbitrary manipulation of multiple objects in 'parallel'.

    Solving the mapping problem, and building a working demo deserves some respect. So you can't think of a decent application after 30 seconds thought, so it must be useless.

    The whole world of manufacturing is based on manipulation of components to produce a more complex assembly. The Universal Manipulator provides new method of achieving this and I am sure lots of applications will be found, because it is easily reprogrammable without retooling and faster than serial manipulation by a single positioning device.

    As an example, PCB populating involves a whole number of pick and place motions to position all the components on the board. If all of the mechanical steps could be replaced by simply dropping all the components on the board and shaking it for a while, assembly would be faster and machines would be cheaper and simpler.

No man is an island if he's on at least one mailing list.

Working...