Engineers have been developing ways to provide robots with a sense of touch for a few decades now. MIT took to the GelSight approach, the University of Texas employed nanowire sensors, and Harvard went with the soft robotics system — all in an effort to provide robots with the ability to ‘feel’ whatever they’re holding.
Engineers from Stanford University, though, have designed an electronic glove that provides robotic hands a sense of touch and a degree of manual dexterity. In a recently released paper, the team showed that their glove was sensitive enough to touch a berry and move a ping-pong ball without crushing them.
The electronic glove is outfitted with an array of capacitors capable of measuring normal and tangential forces. It imitates how human skin layers work together to provide the sense of touch — the outer skin layers on our hands have sensors that can detect heat, pressure and other stimuli. Those layers work in conjunction with a sub-layer of skin called the spinosum, which feature microscopic hills and valleys that amplify that sense of touch.
The electronic glove imitates its human counterpart by providing three-layered sensors on each fingertip that work together. The top and bottom layers are electrically active, while sandwiched between the two are a grid of electrical lines that are positioned perpendicular to one another to create an array of sensing pixels.
A rubber insulator separates the electrical layers, storing electrical energy when the fingers press an object. As that stored energy increases, it provides a way to map the intensity and direction of pressure to specific points on those perpendicular grids, in much the same fashion as human skin.
While the electronic glove is still being developed, the engineers state that robots could wear the touch-sensitive skin for delicate work such as handling eggs on a conveyor belt and putting them in cartons, as well as robot-assisted surgery where precision touch control is vital. Ultimately, the engineer’s end-goal would see the glove automatically apply the right amount of pressure to handle any given object without the need for programming.