Groundbreaking Robotic Arm Can Pick Up Glasses

Robots are ubiquitous today, and some of the most famous ones are already shaped like humans. Of the latter type, a small fraction has human-like hands and “fingers,” which the machines can use to grab a hold of objects, or hull themselves up when they fall to the floor. However, there is currently no design able to modify the intensity with which it grips something, based on the feedback obtained from that object. This is precisely what experts at the Harvard and Yale universities have created.

Their innovation is a robotic arm that features four “fingers” and is able to pick up objects as delicate as wine glasses without damaging them. Unlike other similar instruments, which require complex machinery and software not to damage everything they touch, the simple and soft robotic hand can pick up a host of things, and automatically adjust its fingers so that it has the perfect grip. In addition to the possible applications that the machine may have in various industries and production processes, it could also be used as a next-generation prosthetic device, Technology Review informs.

“When you start to bring robots into human environments, all of a sudden there's a big advantage for being compliant. You can't always know what is where. You don't just want to push through the world and break something. You want to have the mechanics comply,” Georgia Institute of Technology (Georgia Tech) Assistant Professor Charlie Kemp says. The expert is specialized in designing home-assist robots for various applications. Unlike other robotic arms, the new one doesn't have numerous motors and complex electronics.

Rather, it is driven by only a single motor, and has only a minimal array of sensors. But it does stick by a basic type of human behavior, which shows that we do not reach for something using a rigid grip. We keep our fingers relaxed, so as to avoid knocking that object over, and only tighten them once we are sure it is secured within our grasp. The new hand respects this approach, which guarantees that the targeted items will be captured even if minor errors in calculations occur.