Totally agree, this summary really gets the limitations much better. Not at all clear how heads-up poker translates to war games. They must be working on multi player tourney now?
Well to be fair just as they say in real estate*, machine vision needs good lighting, lighting, lighting - just these retina sensors naturally don't like deeply modulated flickering lighting. They like DC lighting, but it needn't be uniform or very bright. We've run this demo in a variety of situations, private homes, lecture halls, offices with lights on and off, shadows across the balancer, outdoors, etc. Some exhibition halls use sodium lighting which flickers very deeply and I guess that was the situation above. I wonder if high frame rate conventional cameras are bothered by this type of lighting also? Maybe they beat against it?
The machine vision community in general doesn't take kindly to an new form of representation that does away with frames, which have been with us since the days of drum scanners. But there's no reason that the beautiful developments of machine vision can't be brought to this new representation, which allows one to think in a new way about vision which seems in some ways closer to how it happens in animals. * location location location
No hoax. There is a rubber cup (what we call the hand) that the tip sits in, but it only serves to let the balancer move the tip laterally. Without the balancer being active, the pencil falls right over. In fact we noticed recently that the performance dropped off (couldn't balance as long as before) and cleaning off all the nice accumulated graphene from the cup helps it hold onto the pencil better. Slow motion recordings with another retina showed that at least once when the pencil fell down it was because noise in the retina input or some other source of noise called for a very rapid movement of the cup that was so quick that the pencil just bounced up out of the cup.
Check out this youtube video for video of Conradt's embedded balancer, which uses only NXP microcontrollers rather than USB + PC. Also the technical paper about this system is here. The page out the silicon retina technology is down right now but will eventually be here.
Slashdot Mirror
Totally agree, this summary really gets the limitations much better. Not at all clear how heads-up poker translates to war games. They must be working on multi player tourney now?
http://spectrum.ieee.org/green-tech/advanced-cars/lithium-batteries-take-to-the-road was written in 2007. I think they got bitten by manufacturing like the early silicon fabs.
These news articles are usually (and this was no exception) short on the root of their woes in manufacturing. Is there any more info on that?
Well to be fair just as they say in real estate*, machine vision needs good lighting, lighting, lighting - just these retina sensors naturally don't like deeply modulated flickering lighting. They like DC lighting, but it needn't be uniform or very bright. We've run this demo in a variety of situations, private homes, lecture halls, offices with lights on and off, shadows across the balancer, outdoors, etc. Some exhibition halls use sodium lighting which flickers very deeply and I guess that was the situation above. I wonder if high frame rate conventional cameras are bothered by this type of lighting also? Maybe they beat against it?
The machine vision community in general doesn't take kindly to an new form of representation that does away with frames, which have been with us since the days of drum scanners. But there's no reason that the beautiful developments of machine vision can't be brought to this new representation, which allows one to think in a new way about vision which seems in some ways closer to how it happens in animals.
* location location location
No hoax. There is a rubber cup (what we call the hand) that the tip sits in, but it only serves to let the balancer move the tip laterally. Without the balancer being active, the pencil falls right over. In fact we noticed recently that the performance dropped off (couldn't balance as long as before) and cleaning off all the nice accumulated graphene from the cup helps it hold onto the pencil better. Slow motion recordings with another retina showed that at least once when the pencil fell down it was because noise in the retina input or some other source of noise called for a very rapid movement of the cup that was so quick that the pencil just bounced up out of the cup.
I guess I don't know how to make a link here to external PDF. The PDF paper from Embedded Computer Vision Conference in Kyoto is here.
Check out this youtube video for video of Conradt's embedded balancer, which uses only NXP microcontrollers rather than USB + PC.
Also the technical paper about this system is here.
The page out the silicon retina technology is down right now but will eventually be here.