Research Video Shows Robotic Hand Crushing an Aluminum Can Like a Human

The most-loved humanoid robots in in pop culture share a common personality type that’s a mix of helpful and chummy. They cook, clean, do laundry, and proceed to high-five you, like the true robotic homie they are. And thanks to a group of researchers, these kinds of machines might not be limited to science-fiction for long.

Meet the robotic hand ADEPT, which is short for Adaptively Driven via Elastomeric Passive Transmissions. It’s 3D-printed, can catch a ball, crush a can, and throw up a shaka.

Kevin O’Brien, an associate professor at Cornell University and the lead author of the research published Wednesday in the journal Science Robotics, tells Inverse the simplistic design could improve the capabilities of robots that already exist in the next few years.

“We initially designed [it] for use in prosthetics, but the possibilities are endless,” O’Brien says. “The technology could be useful in any jointed robotic system from legged robots like Boston Dynamics’ Spot Mini, to improving the strength and sensitivity of Pepper’s hands.

“It’s possible that you could see robots with our technology within [one or two] years.”

While its capabilities to dap you up are unmatched, it’s the materials that comprise and power ADEPT that feel truly game-changing. All of its components are made up of elastomer, a rubbery polymer that feels and acts like human skin when stretched and strained.

Six small electric motors are housed inside of the palm, controlling how it extends and curls its fingers, by winding and unwinding strings much like human tendons. O’Brien and his colleagues dubbed this robot gripping technique elastomeric passive transmissions or EPT for short.

This is complemented by sensors that allow ADEPT to detect the proximity of an object and how tightly it’s being held. This combination allows the hand to nimbly open when it needs to make a quick catch or exert more force when it needs to crush an aluminum can. These kinds of reflexes come naturally to humans, but teaching ADEPT to quickly grasp something took months.

“The most exciting part of the research was the first time the hand used its reflexes to catch a ball,” said O’Brien. “We spent an hour that day having it catch different objects; the simple demonstration was welcome validation for the many months of hard work and difficult engineering.”

This is a further iteration of a long lineage of robot hands and arms. Many of ADEPT’s predecessors are rigid and look more claws than hands. Soft robot hands were proven to be much more flexible, but getting them to react and move like a human is something that O’Brien and his partners pioneered.

Thanks to them, we could one have the Spot Mini play catch with us or get Pepper to throw us a cold one.

Abstract

A new mechanical system has allowed scientists to develop prosthetic hands strong enough to crush a can and reactive enough to catch a ball. The compact and cost-effective technology is a departure from the expensive and clunky motors that control most prosthetic fingers in existence today. The grip strength, grasping speed, and diversity of motions of even the most advanced prosthetic hands pale in comparison with those of a human hand. User studies have shown that 90% of patients with prostheses consider their hand too slow and 79% consider it too heavy. As such, engineering simpler designs for robotic hands without sacrificing adequate precision, force, and speed remains a challenge. Kevin O’Brien and colleagues tackled this problem by creating a cylindrical pulley system comprised of belts wrapped around wheel-shaped gears (often used in motor vehicle mechanics). The resulting cylinders, dubbed elastomeric passive transmissions (or EPTs), could fine tune the grasping force and speed of contact with an object on demand by adjusting the tension in a wire spooled around wheels controlling the cylinders’ movement. The engineers used EPTs to construct an entirely 3-D printed prosthetic hand, which demonstrated a nearly threefold increase in grip force while still maintaining fast finger closing speeds (in seconds) compared to rigid spools. Weighing about as much as a human hand, the prosthetic was able to hold heavy objects such as a wrench. The researchers believe EPTs could be applied to other devices, such as robotic tendons, soft exosuits, and bioinspired mobile robots.