Let’s conduct an experiment, shall we? Look for the closest object you can easily lift with one hand … you see it? Okay, now grab it, pick it up, grip it, and move it around a little (please note: SYFY WIRE cannot be held responsible for any pulled muscles). Nothing to it, right?
Right, that’s because you’ve subconsciously been training your brain since birth about how your hand connects to and manipulates the world around it. It’s second nature, and it’s an amazing product of development and learning. So just think how amazing it would be if you could get a robot to work the same way.
That’s basically the approach of automation technology company Festo’s new BionicSoftHand. But of course there’s nothing basic about it, seeing as it’s inspired by and built to work fundamentally like a human hand, with the innate ability to teach itself, via AI, how to handle and secure objects.
Just look at the gorgeous, muscle-and-tendon-like construction of this thing, comprised of 3D textile knitted fabric, flexible printed circuit boards, pneumatic swivel modules, and fingers made of dual-chambered elastomer bellows …
According to Festo, “the pneumatic robot hand is modeled on the human hand and can learn through artificial intelligence to grasp different objects and turn them into a desired position.” To perform this in-hand manipulation, the BionicSoftHand identifies the parameters of the object, then uses AI to seemingly activate a sense of touch by creating a digital twin, undergoing “massively parallel learning,” and pulling off a “continuous transfer of results.”
The results are pretty astonishing, as you can see in the video above showing the BionicSoftHand dexterously playing with what appears to be a rather large D&D dice.
The BionicSoftHand could obviously have a full of spectrum of useful household applications (check out the YouTube comments for some less useful ones too). And the "soft" approach to robotics makes the human/robot working relationship that much safer.
Yeah, it’s probably safe to say that Baymax, the one and only King of the Soft Robots, would be extremely jealous of this … ahem … handiwork.