Getting to grips with robot hand design
Researchers suggest alternative approaches to the vexing problems of grasping and pinching. Nick Carne reports.
Robots are getting smarter, stronger, sleeker and even more dexterous, but they still can’t peel a potato or button a shirt.
You can consider that one up for humans – temporarily at least – or, like Aude Billard and Danica Kragic, use it as a simple way to define where the immediate challenges in robotics lie.
Writing in the journal Science, Billard, from the Learning Algorithms and Systems Laboratory at École Polytechnique Fédérale de Lausanne in Switzerland, and Kragic, from the Division of Robotics, Perception and Learning at Sweden’s Royal Institute for Technology, examine why designing robotic hands is so difficult, and what the alternatives might be.
“Although research on robot hands has been ongoing for more than five decades, the most common hand used in many applications to date is still a parallel jaw gripper, usually without any extra sensing,” they write.
“Picking up objects with a gripper devoid of sensing is akin to grasping with the tip of your thumb and index finger when both are numb!”
Top of the problem list for hand designers, Billard and Kragic say, is available space. When constructing anthropomorphic robot hands, it is difficult to fit in all the necessary actuators, sensors and mechanical structure.
Then there’s weight. The total has to be low enough to satisfy payload requirements of the arm to which it is attached. That’s why most anthropomorphic hands and prostheses do not have as many controllable degrees of freedom as human hands.
Of course, plastic or metal hands are also more rigid and thus less forgiving of errors or fumbling than ours. It is possible to manufacture softer robot hands, but it requires trade-offs in other areas.
And then there’s skin, which provides the appropriate level of friction and damping and precise information on normal and tangential forces, which is critical for grip adjustment.
“Human skin can also measure stretch and temperature,” the author write.
“By contrast, robot hands typically measure exerted forces through miniature force sensors placed solely at the fingertips.
“Force sensors yield very accurate 3D measurements, but they cannot easily reveal the exact location of contact.”
There are more issues to discuss, but the question has to be asked – and Billard and Kragic ask it – as to whether robot hands have to try to copy human ones in the first place.
“Robotics keeps oscillating between anthropomorphic and traditional industrial designs for hands,” they note.
“But the gripping systems of simpler animals may also provide inspiration.”
In other words, why not create hands that both leverage and go beyond nature?
For example, designers must further ensure, they suggest, that hands are developed in a “plug-and-play” manner and can easily be attached and detached through existing tool-switching systems.
“State-of-the-art force and tactile sensors must become an inherent part of the arm–hand system,” they write.
“Robot dexterity is as much a by-product of advances in hardware as it is of advances in software.”