Slashdot Mirror
Virtual Robots Fooled By Visual Illusions
Roland Piquepaille alerts us to research out of University College London in which virtual robots, trained to "see" as we do, were duped by optical illusions the same way humans are. Here's one of the illusions the software system fell for.
Brought to virtually by Roland and kdawson!
Idiot.
That is just a difference in lighting.
---- Booth was a patriot ----
They could just be programmed to look as if they were falling for it! I smell conspiracy!
622677120
maybe the pixel the robot sampled from each image wasn't identical?
Someone links one of these "Virtual Robots" to a picture of a red dwarf star, that actually points them to goatse's black hole? Now that's an illusion!
Anyone else think this was Richard Nixon's head attached to a robots body?
i'm shocked and amazed.
... but I went ahead and verified with a pixel color id program (ColorPix) that they are the same color.
to being able to upset them with goatse. Maybe that is what starts the robot uprising...
From TFA: "The virtual robots in this study were driven solely by the statistics of their training history and used these statistics as the basis of their correct and subsequent incorrect decisions. Similarly, we believe the human brain generates perceptions of the world in the same way, by encoding the statistical relationships between images and scenes in our past visual experience and uses this as the basis for behaving usefully and consistently towards the sources of visual images." So the robot vision was created as a model of human vision, and it succeeded at doing so. That's sort of interesting, I suppose, but what does it tell us? That we were right about the way human vision works? Seems to me that the point here is really that in some ways, human vision is 'broken' and that maybe it isn't the best apparatus for machines to use. If we want to welcome our robotic overlords, we should be improving on the vision model, not trying to give machines the same flawed framework.
I'm not sure about you guys, but at the moment, I'm kind of doubtful in welcoming our new robotic overlords. I mean, I thought they were supposed to be superior to us and not be fool by petty illusions...
and your eyes do it to. Which is why even sodium vapor lights don't look as yellow as they really are by the human eye. Turn off the white balance on the robots and I bet you they will see them as the same color. Add the average inverse color as a background for each color and your eyes will see them totally different. IE blue behind the orange and orange behind the blue. really stupid test.
OMG Ponies!!! with Glitter!!!! I miss Pink
Goatse is old hat. It's all about 2 Girls 1 Cup these days.
...But you'll want them be easily fooled when they start gathering us for the "Human Re-education Program".
And by "re-education" I mean slaving away in the underground mines.
And by "slaving away" I mean being dumped in the deep hole in the ground.
Mit der Dummheit kämpfen Götter selbst vergebens
show them a Escher staircase.
Cross your eyes, line up the two squares so they're offset by a few millimeters, and then hit the mask. What I saw was that the squares retained their seeming discoloration--until I uncrossed my eyes.
This is why I turn my auto-white-balance off in my digital camera. If I need to adjust the color, I'll do it later in Photoshop. (Another reason to shoot in RAW mode.) -- Carey
To me this is an important step.
It means that we might describe biological vision (ours and most animals) in an more efficient manner. You don't need millions of layered algorithms duplicating evolution. Instead, vision can be described much simpler. We can derive the optimal version of this type of vision and see what holds for biology. We can also try and develop robotics that emulates optimal biological methods and see how well it meshes with our existential experiences of reality. If it meshes well, then you have the foundation for levels of man/machine integration.
This doesn't imply overlords, just a very clean version of robotic humans that could help us take the next steps in our evolution. But, I'm sure that doomsday predictions and irrational fear will come along for the ride as well. Maybe we can explain and correct that as well.
Very cool article. But, now I'm curious how virtual robots will perceive if color blindness was applied...
But then again, being color blind makes a lot of things look the same that shouldn't be...
XML is like violence. If it doesn't solve the problem, use more.
they are also susceptible to the illusion of "beer goggles"? ...Next thing you know, your personal robot's software has it waking up in bed with your new Dyson vacuum and a strange Toaster! [There must be a Bender joke in here somewhere] http://en.wikipedia.org/wiki/Bender_(Futurama)
Look at this one. If you see more clearly for "true similarity" on the red pipe (?), then you will notice slightly a bullshit. :)
Take a look at http://www.lottolab.org/Visual%20Demos/Demo%204.html
The site says the surfaces are "physically identical." I call BS. They are identical only in the sense that they have (assuming this is a faithful rendering of something) the same irradiance per unit solid angle hitting the viewer's eye. They are, in fact, physically different surfaces -- look at the top left corner of each piece, which are facing roughly the same directions and so are similarly lit. The top face is dark and lit more brightly, and the bottom face is light and lit dimly, and a robot that observes notices this (like humans do) is pretty impressive.
Demo 15 (the one in the article) is a bit more legitimate, but it's hard to tell whether the left disc is redder or is just lit with red light.
Since we are not born with flashlights glued to our heads (although in Soviet Russia one can be obtained by getting into a fistfight), we must compensate for the tone of ambient lighting. This correction that we easily do in our heads but must be applied manually on digital cameras in fact allows us to determine true color of the objects more accurately in natural settings. Therefore I wouldn't call this an optical illusion any more than the fact that our eyes become more sensitive at night.
... is a virtual robot? just some software? then say so.
no, i haven't read the article.
...that's a difference in surrounding lightning.
...WHEN those colors are ISOLATED. (i.e: putting the name "red" "orange" "purple" on a color you see alone).
Human visual system (as most other senses) work not by absolute values (i.e.: it doesn't see that the color '#c0ff20' or whatever), but mainly by comparing the signal with signals from the surrounds.
Thus what we technically see is that on the left object the central case looks much more "greener" than its surrounding, in the right object, the central case is "much more orange" than the surrounding. In fact, when the mask is enable, the colours do change from the point of view of the visual system : we were seeing contrast with two different surrounding, now we see a contrast with a third surround (mostly black). We see three different contrasts, even if from the computer's point of view the color is them same (the same RGB triplet / same intensity on your CRT/LCD)
If the scientist are trying to build efficient visual systems, they are probably mimicking this "works-by-comparing" method that the nature is using.
That's why we can recognise the same object, during day, during night, with weird lights, displayed on the screen (worse colour gamut) or on a print out (even worse color range). Because the relative difference stay the same, even if the colour as-seen-by-a-computer change.
The same is valid for any other sens, or in fact, any other information that is processed by neurons. Everything works by comparing (across several signals, across time, etc.). There's no such thing as "an absolute value" in the information carried by neurons.
That's also why all those "but the human eye can only x thousands of colors" (usually mocking the latest 32bit, 48bit, floating point or whatever color depth), are fundamentally wrong.
Yes, the human visual system can only distinguish a hundred or so colors....
When two colors are put next to each other, the human brain can suddenly distinguish much more subtle variations (each color would be considered as "brown" when seen alone, but next to each other, you can use thousands of different shade of brown and the eye will still see the difference).
That's also why radiologist are fond of high contrast / big depth screens : because all those difference in shades of grey *can* be distinguished and *are* revellent for the diagnosis when displaying X-Ray pictures.
"Sufficiently advanced satire is indistinguishable from reality." - [Tips: 1DrYakQDKCQ6y52z6QbnkxHXAocMZJE61o ]
I'm struggling to find the utility of the study. So, if we learned to see differently, we could see the world in a way different enough to not be fooled by certain optical illusions, and probably be fooled by others?
Assuming it is possible to change the way a human sees without breaking the brain. A popular theory on evolution is that we evolved our brains to better analyze visual data coming in. We're not deceived as easily by certain camouflages animals use. Stripes, dots, color, etc.
Confirms what we thought about the way we learn to see, perhaps? That'd make sense.
I'm not colour-blind (yes, I've done the tests, I'm sure) and the colours look the same to me too. Well maybe not identical, but pretty close, and the right-hand one certainly doesn't look orange to me like it's supposed to.
Do you think these robots can be programmed to see the little dolphins swimming into the sunset of a Magic Eye book?
Arrrrgh! /Mallrats reference.
Sorry about the writing. Robot fingers, you know? Cliff Steele in DOOM PATROL #23
Don't give that asshat any hits!
Did the left hand square appear orange to you, and the right side green?
I'm normally affected by optical illusions just like everyone, but I don't see it here - the center squares of both discs appear equally light green to me. Any ideas? I don't think I'm colorblind...
....."It was amazing, They done exactly what we programed them to do. Next we are going to program one to *not* be fooled by optical illusions and see what happens! I can't wait!"
God Be Gone
As a designer of SCADA and control systems, I like to tell people that "if you can see it or do it, then I can measure and control it."
Of course, that's not entirely correct. The concept of building machines or robots that exceed our capabilities is something that interests me. It reminds me of this previous article, and I wonder what kinds of things future machines will be able to "sense" and "do" if they themselves are built from machines with enhanced capabilities.
For now, though, it's nice to know that they are inherently limited to our own capabilities as humans.
Adapt, adopt, or get out of the way!
apparently no one thought of making red-green color blind robots; as a somewhat red-green color blind human, those greens still look the same to me ... go figure.
don't they look rather... shapely?
or maybe thats from watching too much porn...
Whatever it was. I've pretty much thrown up my hands at trying to say something 'looks' green or orange or anything if it is in the realm I can't discern, because I recognize I simply have no comparative base to describe what something looks like to me in order to match a normal color vision person's perception.
However, I do feel confident that it's accurately characterized in my case as a dramatic insensitivity to red, so it is a decent bet that to me that everything looks less red (i.e. brown looks green is my logical guess, but no way of knowing I map green to the same thing other people map green to, etc.)
XML is like violence. If it doesn't solve the problem, use more.
you don't want to know
-Clio
Karma: Bad (mostly from not giving a fuck)
Blog: http://clintjcl.wordpress.com
In the virtual world I created, Smorgons are 6 meters tall, shoot acid out of their noses, and have been known to breed 10 offspring in a month.
In the real work, however Smorgons don't exist, so therefore I must conclude that virtual tells us nothing about actual.
HexaByte - he's a square and a half!
Ok, so what if your robot is fooled by some obscure optical illusions-- other 'illusions' (or for lack of a better term, optical phenomena) are far more problematic- consider the problem of recognizing that you're looking at a mirror and not just a big room. Or the problem of 'seeing' the subtle reflections cast by a transparent medium like a window, in order to recognize the presence of an obstacle. Speaking as someone who's done a fair amount of work on autonomous robot exploration, these are big unsolved problems for robots equipped with off-the-shelf cameras.
So long, and thanks for all the Phish
I wonder if you're getting a kind of Persistence Of Color effect.
Here's one my favs:
http://www.johnsadowski.com/big_spanish_castle.php
Cube On! (http://stores.ebay.com/PuzzleProz)
If we ever build AI that's able to perform the kind of "intelligent thought" that we humans are capable of, then this is exactly the kind of thing we should expect to run into more and more. I've always contended that the closer we come to building computer circuits that mimic the processing within the human brain, the less that output will be what we generally consider to be computer-like (logical, predictable, mathematical).
This one is beyond me. It's just two different spheres represented in two dimentions with diferent coloring. What's the big deal? I cannot see any optical illusion and I can't see how a computer would not detect the complete differences in pixel encoding. Maybe I'm too stupid to understand, but I see no optical illusion and I can't imagine how any computer program could think there is one if they are comparing ones and zeros.
Maybe I'm missing some higher level context. Please explain what this is about. It's just two pictures which are completely diferent to my comuter.
I wonder how they'd respond to this (http://www.planetperplex.com/img/fake_illusion.jpg) optical illusion.
Here is another example that I think is better http://img107.imageshack.us/my.php?image=167ta7.jpg
This is caused because the brain processes details about lighting, color, illumination, etc. Otherwise you'd see a piece of white paper with a shadow on it and think that the piece was half black and half gray. I don't remember what this phenomenon is called (I thought it was color persistence, but it apparently isn't). Also, most insects/small animals lack it, so they would perceive the same object under different lighting as a different object.
Send email from the afterlife! Write your e-will at Dead Man's Switch.
I'm working on the phenomenology of vision in my MFA thesis, and it occurs to me that no one is being fooled. You and the robots, are correctly perceiving the color balance caused by color of light of the form. And while those two colors may be the same in a 2-d image. In the real world if you saw the one color in red-orange light and the same color in blue light, they would in fact be very different colors. I auto-white-balanced each section to reveal how the colors would look with out being in a tinted light. And Voila, it is exactly as you probably (correctly) perceived them, green & orange. Cheers.
That's not true....I took a screen print and imported into paintbrush and the colors are different one is tan one is green.
http://portal.acm.org/citation.cfm?id=1080441&dl=GUIDE&coll=GUIDE
Artificial intelligence is the study of how to make real computers act like the ones in the movies.