World's Fastest Robot Versus the Wiimote 92
kkleiner writes "Adept's Quattro, a placement and sorting arm, took the title of fastest robot last year, but it was only recently during National Robotics Week that it met its most gruesome opponents: nerds with Wiimotes. Visitors tried to keep the Quattro from placing and sorting on a small mechanized platform by moving it using the Nintendo video controller. The bottom line is that when it comes to simplified and repetitive tasks there's really no beating robotic prowess."
Extra Extra: Robot Beats Nerds! (Score:2)
Oh I don't know about that... (Score:1)
The bottom line is that when it comes to simplified and repetitive tasks there's really no beating robotic prowess
It all depends on the simple, repetitive task.
Re: (Score:2)
Yeah. I bet I could beat that robot at Breathing!
Re: (Score:1)
Re: (Score:2)
... you would be typing
Frist post (Score:2)
Quattro? (Score:5, Funny)
Re: (Score:2)
I think of that every time I see a Flexpicker or Quattro.
Oddly named because it has 4 arms, and I always think they should've followed the Schick Corporation's lead and skipped from Flexpicker's 3 straight to 5.
Robotics improving our lives (Score:2)
The bottom line is that when it comes to simplified and repetitive tasks there's really no beating robotic prowess.
Now if they can just teach the robot to play MMORPGs!
Re: (Score:2)
That happened first, and years ago. There are automation scripts so sophisticated they can kill 10 mobs and /shout "STFU n00b!!!!111elebenty-one" every 30 seconds.
Really, virtual environments get solved first, and most easily. The real world is a good deal harder to deal with because analog data is noisy, incomplete, occasionally bogus, and often misleading. Digital data from a virtual environment is perfect by default. Robotics developers trying to
Re: (Score:2)
There are automation scripts so sophisticated they can kill 10 mobs and /shout "STFU n00b!!!!111elebenty-one" every 30 seconds.
That's not particularly sophisticated, though probably enough to buy them a bit more time once they're caught.
virtual environments get solved first, and most easily.
Well, in a sense, yes. They also make things similarly easy for the defense -- for instance, while it's easy for an aimbot to headshot people across the map and through obstacles, it's also easy for the server to log the entire event, and then perform some datamining, looking for how accurate the bot is.
Re: (Score:2)
Someone could write an aimbot that simulates a human-level of inaccuracy. The server would have a hard time telling the difference between a bot and a player with a really good aim, and you don't want to accidentally ban a legitimate player.
Re: (Score:1)
Re: (Score:1)
Re: (Score:2)
Now if they can just teach the robot to play MMORPGs!
I think they already taught it simplified and repetitive tasks.
So wait (Score:4, Funny)
You took a robot, capable of crunching numbers at speeds in excess of a thousand calculations per second, programmed it and engineered it to perform a specific task, and then wanted to see if humans, who take 1/5th of a second just to react, can't do any more than a few SIMPLE calculations in a second, and had them use the worlds laggiest controller, and wanted to see who would win?
Is this like, one of those Hypothesises that's bound to be true by the laws of physics, but you gotta test it anyways?
Re: (Score:2)
Is this like, one of those Hypothesises that's bound to be true by the laws of physics, but you gotta test it anyways?
Mythbusters influence strikes again.
Re: (Score:1)
But you have to admit, it looks pretty darn cool.
Re: (Score:2)
Re:So wait (Score:4, Interesting)
It's a pick-and-place machine. Most PnP require that the inputs and outputs are stored in well-known locations, and have pretty basic image recognition software (they can tell if a black blob is in the wrong place, for example - if it was loaded wrong). Or to handle the slight misalignment of the source or destination.
In this case, the robot is picking and placing from and to a platform that can move arbitrarily, while it's even doing the picking and placing. That implies it not only knows it has to look for the source and destination, but recognize the platform and perform the task. Even if the thing it's grabbing suddenly decided to move under it while it's doing the picking or placing.
The human might be slower, but they're also a lot more unpredictable, so the robot has it use up its millions of calculations per second to figure out where things are and react when things start moving from under it.
Re: (Score:1)
If you watch the video you'll see that the destination platform is under computer control. The pick and place machine probably knows (well in advance) where the wiimote commands have told the platform to move. There is a 100-500ms latency between tilting the controller and the platform velocity changing.
If the platform was attached to a long handle that a human could pull back and forth quickly I'd be more impressed. The platform could have a sensor underneath to tell the robot where it was or you could mak
Re:So wait (Score:4, Informative)
It's a pick-and-place machine. Most PnP require that the inputs and outputs are stored in well-known locations, and have pretty basic image recognition software (they can tell if a black blob is in the wrong place, for example - if it was loaded wrong). Or to handle the slight misalignment of the source or destination.
In this case, the robot is picking and placing from and to a platform that can move arbitrarily, while it's even doing the picking and placing. That implies it not only knows it has to look for the source and destination, but recognize the platform and perform the task. Even if the thing it's grabbing suddenly decided to move under it while it's doing the picking or placing.
The human might be slower, but they're also a lot more unpredictable, so the robot has it use up its millions of calculations per second to figure out where things are and react when things start moving from under it.
I'm almost certain that the inputs and outputs ARE in well-known locations. As a robotics engineer, the first thing I noticed on looking at the video is that the movable part is mounted on a very solid, rigid, linear actuator. That thing knows the location of the plate to within microns. The second thing I noticed is that the plate that "moves arbitrarily" moves very smoothly with slow acceleration.
So you have a high speed robot putting things on a very slowly moving (compared to the actuator speed) plate, the position of which is known precisely. It would be impressive if the plate could move in 2D or 3D and had a handle for people to move it around with, but as it is.. not impressed at all.
Re: (Score:3, Interesting)
Ah grasshopper you have achieved the first level... But to reach zen you must go to the next level. And after that one day you might even reach the following:
http://www.youtube.com/watch?v=xNG3sgk02Lc [youtube.com]
http://www.youtube.com/watch?v=q395-F6hAcg [youtube.com]
I want a machine to beat that! My head just spins thinking, ok so where did that cup move to?
Re: (Score:2)
This kid is faster: http://www.youtube.com/watch?v=c5H04yGfklY [youtube.com]
He can stack faster than many people can finish pronouncing his name ;).
That said, you can probably build a robot to beat that. Robots have been faster and more precise than humans for years.
The big thing about the robot in the article is it can "stack" on stuff that's moving _arbitrarily_.
Re: (Score:3, Funny)
Brain the size of a planet and they have em doing simple sorting.
Re: (Score:1)
I'm sure the robot would still have been able to deal with that to. Would have been nice to see.
Re: (Score:1)
Let him pick small fuzzy things on a fuzzy surface. Hankerchiefs,
reminds me of Aliens (Score:5, Funny)
They're pretty good at working on humans, too (Score:3, Interesting)
See:
http://en.wikipedia.org/wiki/Robotic_surgery [wikipedia.org]
When my mother had her hip replaced, the surgeon said that robot-replaced hips lasted longer due to better hole size and placement, (they make a hole in your bones and hammer the replacement joint in...)
As for this kinda flaky 'robots vs. humans' story:
1. We'll never be able to beat a robot's reaction times {see note} speed and/or raw power, but
2. Until AI improves, we'll still be the ones programming the things
Note: What was that SciFi story about humans being 'paired' with cats in order to have both high intelligence and inhumanly fast reaction times?
Re: (Score:2)
Re: (Score:2)
Kinda depends on why the robot fucked up, doesn't it?
If it wasn't used properly, the human surgeon/operator.
If it had a manufacturing defect or programming error, the manufacturer.
Re: (Score:2)
Neither. You just sue the hospital/clinic - they, or their lawyers and/or insurers will take care of passing off the blame - I mean identifying the guilty party - for you. Probably ever faster than a robot :)
Re: (Score:2)
If it wasn't used properly, the human surgeon/operator.
If it had a manufacturing defect or programming error, the manufacturer.
What about if there's a power cut? Should a UPS be installed on every critical robot, together with a requirement for a backup generator?
Re: (Score:2)
Uh... yeah, I'd say so. Isn't that pretty common for hospital equipment?
If the hospital/clinic/whatever fails to take reasonable precautions against power failures during surgery I'd say they're probably going to be held liable for the consequences.
Re: (Score:1)
Re: (Score:1)
A robot playing tennis? Anyone
Re: (Score:2)
Sorry to undo the moderation of my AC friend, but The Game of Rat and Dragon [gutenberg.org] is available from Project Gutenberg!
Oblig. (Score:1)
Nerd w/ Wiimote: It doesn't look so shiny to me.
Quattro: Shinier than yours, meatbag.
Re: (Score:2)
really impressive (Score:3, Interesting)
Go to the linked article (yes, yes, do it anyway). Skip the Wii demo video that forms the basis of the post because it really isn't interesting. Go to the second video. Watch it.
Holy frick. Robotic vision and control has come a long way.
Re: (Score:3, Interesting)
Yeah, but am I the only one here who thinks that a lot of those tasks could have been done with machinery orders of magnitude simpler and cheaper than this robot? Seriously, half their examples of 'real world' usage were moving things from one conveyor to the other, with no sorting or filtering required. Some of their examples (like placing the chocolates in the correct locations in the box) were impressive, but it just felt to me like they were showing off when much simpler designs could have been used.
Re:really impressive (Score:5, Insightful)
I wonder if one of the implicit advantages of a highly flexible, programmable robotic system like this, rather than special-purpose hardware, is manufacturing flexibility.
I know that chocolate manufacturers need to retool their lines quite frequenty (Valentine's Day, Easter, etc.), and imagine that's true for lots of industries. Many of the examples from the second video are food handling: a processing plant that handles frozen burgers one week might be making chicken nuggets or fish sticks the next.
Re: (Score:2)
Companies pay more money for flexibility. Food manufacturing in particular is one industry where the flexibility offered by vision-guided robotics provides an overall cost advantage vs. multiple automation systems for specific products.
IIRC, Adept specifically markets this robot to the food, medical, and semiconductor industries because it is cleanroom & washdown rated. Because the servomotors and electronics are all contained within the box at the top, it's much easier to keep the g
Re: (Score:2)
Oh? No sorting or filtering?
Then I suggest you look again at the video. What I was seeing was random placement of items on the source conveyor belt and quite orderly placement on the destination conveyor belt.
Re: (Score:2)
I took him to mean 'all the items are the same type.' As opposed to, say, having to assemble a sandwich from a random assortment of ingredients coming down the line.
Re: (Score:1)
Yeah, but am I the only one here who thinks that a lot of those tasks could have been done with machinery orders of magnitude simpler and cheaper than this robot? Seriously, half their examples of 'real world' usage were moving things from one conveyor to the other, with no sorting or filtering required. Some of their examples (like placing the chocolates in the correct locations in the box) were impressive, but it just felt to me like they were showing off when much simpler designs could have been used.
as far as I can tell the other video is a promo video, not real world applications.
Currently most of the tasks show are done (faster and at a considerably lower price) by a high speed conveyor belt, a couple of metal fences, a little physics and an electric eye to sort the backward from the forward.
The machines ability to quickly find and organize is impressive.
The inability of a person to out maneuver the machine using a mechanized tray on a linear track, moving at a fixed speed that is slower than the
So will manufacturing return? (Score:2, Insightful)
Re: (Score:3, Interesting)
Whenever I see these reports, it makes me wonder about the implications on manufacturing. Someone in the US or Europe can't/won't compete with someone in China working 15 hour days in a sweatshop for 50 cents an hour, and so from the company's standpoint, it makes economic sense to move. But will the rise in robotics cause a return of manufacturing? You will still need some people working in the factory maintaining the robots and what-not, but it may be cheaper to manufacture things closer to their destination rather than manufacturing them in a developing country and shipping them.
You do realize that manufacturing in the US has been automated for a very very long time and it is *still* going down the drain.
But even with automated manufacturing you still need manual labour to work the production lines. Its economically infeasible to produce a robot that has the flexibility and dexterity of a human for general purpose use. You should watch the shows like "how its made" etc. They show lots of automated processes, yet there are always manual steps involved. I remember seeing one s
Re: (Score:2)
You can pay for a lot of coolie labor for the price of one good robot, and with far less up front investment.
Re: (Score:3, Insightful)
And no unions, and no insurance . . .
Not saying its a good thing, but a lot of companies would gladly take a robot over a human any day, just to avoid these two.
Re: (Score:2)
Sure, but they don't need to worry about that in China.
Re: (Score:2)
The rise of robotics has caused the return of manufacturing. China has been losing manufacturing jobs for quite some time now, as they lose jobs to robots faster than America ships jobs to China. Don't assume that just because there a fewer manufacturing jobs in America each year, that there is less manufacturing capacity. Most broad measures of industrial capacity show a 50% increase over the past 25 years, with "autos assembled" being the main exception (and those were basically flat until 2 years ago)
Re: (Score:2)
Re: (Score:2)
The Fed tracks various measures of industrial production and capacity. There's a wealth of raw data here to dig through. http://www.federalreserve.gov/releases/g17/download.htm [federalreserve.gov]
It has been this way for a while. (Score:3, Interesting)
It's been known for a long time that robots and computerized systems are vastly superior to humans at simple tasks, their only downside is the upfront cost and often inflexibility.
One of the neatest applications I saw recently was in a factory where macadamia nuts were shelled. The nuts would pass through a big set of rollers, cracking the shells open. Then, the shell casings and the nuts would fall down, and a computer vision system would detect the nuts and the shells. Everything then fell through a collection of compressed air blowers, that would precisely blow the macademia nuts out of the stream of falling shells onto a conveyor platform, while the shells would fall seperated into a hopper off to somewhere else.
Re: (Score:2)
Indeed.. it's why one Dutch entrepreneur is still shipping tons and tons of onions off to Poland to be graded, cut, peeled, and cut further (for rings, dices, etc.) by hand by 'low wage' workers, and then sent back to NL, rather than investing in something like...
http://www.onionpeeler.com/item.asp?id=3276&/NAKAYA/Automatic_Onion_Skin_Peeler_%5BNon_Abrasive%5D/ [onionpeeler.com] ...which only does the peeling part for regularly-shaped onions (at least it take
a linear test only (Score:2)
This is equivalent to testing how well a machine can put parts onto a variable speed conveyor belt. It is industrially useful, but not particularly interesting.
TFA's claims that 'the competition in the video above would have been closer if the platform had moved faster, but then you’re really talking about machine vs. machine.'
We are already talking about a machine vs. a machine. Pitting a specialized machine against a more basic machine will generally have one outcome. As the response speed of the
Yay for nostalgia (Score:2)
The good news? Nintendo is releasing Stack Up for the Wii.
The bad news? R.O.B. will cost around $50k.
I am so going to have fun when they release Gyromite. When one of those suckers gets loose it won't just destory your TV, it'll take out your entire living room.
Human speed? (Score:1)
Re: (Score:2)
The human could build a better robot, and use that to defeat the other robot at repetitive tasks :)
Not a fair comparison (Score:1)
Apparently the robot has access to the position of the platform. It would only be really impressive if it was using computer vision to see where a platform is. All this proves is that it can do some simple math and move really fast.
Re: (Score:2)
What makes you think it's not being done with machine vision?
Interesting, but (Score:2)
Most of the packaging tasks that this machine performs are done by third-world workers who earn less than $5000 a year. Especially technology items: your new Microsoft mouse or iPod was packaged in China by workers who live at the factory and work long, long hours
That fancy robotic pick & place machine is impressive and it's much faster than a human. But it's not faster than a bunch of humans and when those humans are Chinese they cost much, much less than the robot (and its custom "workplace") does. T
Re: (Score:2)
Re: (Score:1)
O RLY? (Score:1)
And Complex Tasks Also (Score:2)
Robots can also accomplish quite complicated tasks as well. I've built robots that could change the tools and fixtures that they depend upon and cross feed conveyors which were also robots of a sort. When these units have computerised saws, lathes and mills as well as doing assembly the only issue is economic viability due to the skilled employees needed to modify and service the systems. But a wide variety of product can be quickly produced. In essence this is the American approach to robotics in ind
That's nothing. See this high speed robot hand. (Score:5, Informative)
That's nothing. The Adept robot is in production. Here's what's working in the lab. [youtube.com] Watch the fingered robot hand tie knots in a rope, dribble balls, and throw a cell phone in the air and catch it in a different grip, all at about 5x human speed or better. This system has 1ms visual reaction time.
Working at very high speed has advantages. Once the reaction time of the systems is faster than movement caused by gravity and other disturbances, flexible objects like ropes and cloth can be manipulated in a straightforward way.
Re: (Score:1)
Hilarious... (Score:1)
Obituary for human menial labor is premature. (Score:2)
A robot does fine at packing uniform objects into uniform packages, but good luck finding one that can pack your shipment from Amazon.com, or do basic construction work, or pull the ingredients for 50,000 gallons of Coca Cola off of shelves in a warehouse and mix them all together for you. There is still a lot of dull, brain-killing, menial work for humans to do. If you work in an office and have never set foot in any kind of industrial operation, you'll probably be surprised at how much stuff still needs t
Re: (Score:2)
Re: (Score:1)
A robot does fine at packing uniform objects into uniform packages, but good luck finding one that can pack your shipment from Amazon.com, or do basic construction work, or pull the ingredients for 50,000 gallons of Coca Cola off of shelves in a warehouse and mix them all together for you. There is still a lot of dull, brain-killing, menial work for humans to do. If you work in an office and have never set foot in any kind of industrial operation, you'll probably be surprised at how much stuff still needs to be done by humans.
There are already automation systems for warehouses, for instance robotic forklifts http://www.inro.co.nz/ [inro.co.nz]
Humans are great at packing odd shaped stuff. But when you have a lot of regular objects, look out for robots. This is just a matter of low hanging fruit - why make a flexible robot when simple robots are cheaper and the market is still huge and unsatisfied.
Grabbing large quantites of ingredients is a large scale logistical exercise. There are very few "moves" and a lot of small difficulties. Humans mak
hmm (Score:2)
Lucy could not keep up with that. (Score:1)
I am impressed. As far as the reference to Lucy goes, think of the episode where she has taken a job in a candy factory.
It's not the human it's the linear actuator. (Score:2)
Man, this is old. (Score:1)
Where's the ghost of John Henry? (Score:2)
I Love Lucy (Score:1)
deciving app (Score:1)
The specific application could, however, be implemented by using even slow robot. The trick is to use what's called conveyor tracking. The button panel is connected to encoder that is synchronized with robot coordinate system.
Usually a more complex version of the application is used: the targe
I would have liked... (Score:1)
...to have seen the robot vs a person moving the tray. This wasn't robot vs human. This was robot vs wiimote lag. The video only serves to demonstrate my biggest problem with the Wii as a platform - it's far too imprecise for anything that doesn't require flailing the remote like a spaz.