If you’ve ever looked at the odd images formed inside Jell-O when you thrust in your spoon, you’ll understand GelSight. The idea was to design for robots working with complex objects an unusual and very powerful fingertip sensor similar to our own. But the MIT team has since realized that their creation has equally exciting applications elsewhere.
GelSight is essentially a piece of soft optically transparent gel with one shiny face as a sensor–when it’s pressed onto a 3-D object, the sensor surface distorts to match the object’s shape and some cameras peeping at the reflections of the twisted surface can “see” what the gel is touching. The cameras are tighly focused on the reflective surface, and their image feed is processed by a computer to generate a 3-D image of the thing being touched.
It’s actually powerful enough to detect precisely the raised ink printed on a $20 bill–and actually far more detailed information than this, right down to objects that are a micrometer in depth and two across. A typical red blood cell is seven micrometers across.
Our technology for creating the right kind of resilient, transparent gel is very well advanced, and our digital imaging tech and 3-D image processing is so advanced that a 3-D compact consumer camera is just a few hundred dollars: Combining these two advances into one has resulted in something that’s surprisingly powerful. The trick was to perfect GelSight’s reflective surfaceby refining the tiny flecks of metal fixed to it in a type of flexible paint, optimizing the lighting of this surface inside the gel. The feat is still being refined.
Previously 3-D imaging in this type of microscopic detail required large expensive gear, like a confocal microscope which may even have required fitting onto a vibration-damping optical table consisting of a granite slab and active damping legs. Now the GelSight camera is the size of a Coke can, and can be dabbed onto almost any object by hand–producing 3-D images instantly.
Such 3-D images could transform how your dermatologist tracks the development of skin moles, looking for signs of melanoma, and it could enable almost instant identification of bullet casings at the scene of a crime, and what it could do for instant biometric security is almost incalculable.