MIT develops a portable, high-resolution 3D imaging solution | Hacker News Reader

MIT develops a portable, high-resolution 3D imaging solution

bostinnovation.com

88 pointsdmcgregor10 comments

I wish HN had a "wow" button. Simply upvoting a submission like this one seems... inadequate.

Context: GelSight can register “physical features less than a micrometer in depth and about two micrometers across,” which means it can create a 3D image of ink on a piece of paper (below):

Below that there is a 3D image of ink on a piece of paper. Taken with a handheld device. "Wow" about covers it, although I might go with "WOW."

That's an awesome idea. Perhaps it could kick in at a karma level.

So has everyone else, apparently. (I'm at Siggraph. (Off topic, I also learned I can't really see glasses-free 3D, but I'm not stereo blind. One eye may be slightly more dominant, so it could be an issue of 'calibration' I suppose.))

MIT's is definitely the coolest and best I've seen though, I don't think I've seen anything like it. Everyone else's is a light-based scanner or algorithms to 3D-ify mostly flat imaging. This amazes me beyond no end.

"When the metallic side is pressed against the surface of an object, the rubber deforms slightly"

while the concept is great, it still doesn't maximize precision (for high resolution analysis), since the pressure can also deform your object of interest. ideally, you would want to use a metallic as less viscous as possible to keep the surfaces rigid.

This is truly remarkable. I woke up in the middle of the night after reading this with a few improvements that can turn this into a human-equivalent level skin sensor for robotics.

Temperature: add something that changes color based on temperature.

Wetness: add some microchannels and watch them as they wick moisture into the material

Pain: introduce small ink packages that will burst if the material is damaged.

Vibration: add a microphone to detect pressure changes in the touched object, or if your cameras are fast enough just watch the surface as it moves up and down.

This could also be simply expanded to senses we don't have: Magnetism (ferrous particles), radiation(scintillation), PH level(litmus), detecting the presence of any chemical directly(specialised receptor), etc... this is revolutionary.

Using cameras behind a gel is a simple, non-obvious(to me), solution to the contact sensing problem. This is going to be used everywhere.

I wonder how easy a cheap, low tech version of this would be to make. A camera staring at the underside of a clear gel with an incredibly thin layer of opaque reflective pigment on top of it could pick up and translate a depth map if the surface were evenly lit.

The computer vision part might be related to polynomial texture mapping: http://www.hpl.hp.com/research/ptm/

I don't understand how this works. Little help hn!

Inside sources say that it will be called Nitendo 3DS

Reply on news.ycombinator.com