Robots with feelings - in their fingers

Robots now have the sense of touch, thanks to University of Southern California researchers Jeremy Fishel and Gerald Loeb. The robotic fingers are modeled after our own -- with a “bone”, flexible skin with fingerprints and fingernails, and a liquid filling to mimic finger pads. Using their BioTac® sensors, robots detect and correctly identify a wide array of natural surfaces, even outperforming humans. The team also developed a novel decision-making algorithm, dubbed Bayesian exploration, which uses prior knowledge to decide how to touch a surface to correctly identify it.  The tactile sensors allow the robots to determine an object’s texture, direction and location of applied force, and temperature. Standing to revolutionize human prostheses, the sensors may also be applied to personal assistive robots and testing the feeling of consumer products, although the robots have yet to determine what textures humans prefer.