Multitask Like Never Before With These Robotic Fingers

There are several explanations for why the human hand developed the way it has. Some researchers link our opposable thumbs to our ancestors’ need to club and hurl objects at enemies or throw a punch, while others say that a unique gene enhancer (a group of proteins in DNA that activate certain genes) is what led to our anatomy. But most agree that bipedalism, enlarged brains and the need to use tools are what did the trick.

Yet, for as dexterous as our hands make us, a team of researchers at the Massachusetts Institute of Technology think we can do better. Harry Asada, a professor of engineering, has developed a wrist-worn robot that will allow a person to peel a banana or open a bottle one-handed.

Together with graduate student Faye Wu, Asada built a pair of robotic fingers that track, mimic and assist a person’s own five digits. The two extra appendages, which look like elongated plastic pointer fingers, attach to a wrist cuff and extend alongside the thumb and pinkie. The apparatus connects to a sensor-laden glove, which measures how a person’s fingers bend and move. An algorithm crunches that movement data and translates it into actions for each robotic finger.

The robot takes a lesson from the way our own five digits move. One control signal from the brain activates groups of muscles in the hand. This synergy, Wu explains in a video demonstration, is much more efficient than sending signals to individual muscles.

In order to map how the extra fingers would move, Wu attached the device to her wrist and began grabbing objects throughout the lab. With each test, she manually positioned the robot fingers onto an object in a way that would be most helpful—for example, steadying a soda bottle while she used her hand to untwist the top. In each instance, she recorded the angles of both her own fingers and those of her robot counterpart.

Wu used that data to establish a set of grip patterns for the robot and a control algorithm that would provide the correct assistance based on a given hand position. 

7 Finger Robot

Watch on

While the robot, which is only a prototype, can change its position, it can’t yet mimic the force or grip strength of a human hand. “There are other things that make a good, stable grasp,” Wu told MIT News. “With an object that looks small but is heavy, or is slippery, the posture would be the same, but the force would be different, so how would it adapt to that?” The team isn't dicussing how it plans to measure and translate force yet. 

Machine learning, or the ability of a computer to adapt its processes based on data, could allow the system to adjust to the preferences of a given user. Wu says she could pre-program a library of gestures into the robot. As someone uses it, the robot would sync up with how a person grips objects—not everyone peels an orange the same way, right?—and discard grip types that aren’t commonly used.

Asada also says that the device, now rather bulky, could eventually be made foldable and one-third of its current size. He envisions a watch with robotic digits that appear and retract when needed.

While Asada and Wu see the utility of their robot for persons with disabilities, it’s also part of a larger robotics movement that seeks to endow able-bodied users with super-human characteristics. Another MIT system, for example, works on the same principle as Wu’s robot, but adds extra arms instead of fingers, allowing wearers to open doors with their hands full or hold an object steady while hammering.

For the most part, these wearable robots are about adding strength. The TitanArm, developed by students at the University of Pennsylvania, allows its wearer to lift an extra 40 pounds. More ambitious setups involve full exoskeletons that inch ever closer to Iron Man. For instance, Daewoo Shipbuilding and Marine Engineering, a South Korean company, has outfitted shipyard workers with suits that allow them to hoist slabs of metal and lumber with relatively minimal effort.

What all these approaches have in common is how simple they are to use. Users don’t need to learn control schemes to manipulate their robotic appendages, but instead go about their tasks, relying on an animatronic spotter to help them on their way.

Corinne Iozzio | | READ MORE

Corinne Iozzio is a New York–based technology writer and editor. When she’s not fiddling with LEGOs or Nerf blasters, she covers gadgets and emerging tech for various publications, including Popular Science and Scientific American.