Robotic Hand for Amputees: Real-Time Tactile Control
For more than 5 million people worldwide who have had an upper limb amputation, prosthetics have come a long way. Beyond traditional mannequin-like appendages, a growing number of commercial neuroprostheses are on the market—highly articulated bionic limbs designed to read the residual muscle signals of the patient and robotically mimic their intended movements.
However, this high-tech dexterity comes at a cost. Neuroprosthetics can cost tens of thousands of dollars and are built on metal frames with electric motors, which can be heavy and rigid.
Engineers at the Massachusetts Institute of Technology (MIT) and Shanghai Jiao Tong University developed a soft, lightweight and potentially inexpensive neuroprosthetic hand. People with amputated limbs who tested the prosthesis performed everyday actions—zipping up a suitcase, pouring juice from a carton and petting a cat.
They managed all of these just as well, and in some cases even better, than people with more rigid neuroprostheses. The researchers found that the prosthesis, equipped with a tactile feedback system, restored some primitive sensations in the volunteer's residual limb.
The new design is also remarkably durable, quickly recovering from a hammer blow or being run over by a car. The smart hand is soft and elastic, weighing about 200 grams. Its components cost around $500. That is a small fraction of the weight and material cost of more rigid smart limbs.
“It is not a product yet, but its performance is already comparable to or better than existing neuroprostheses, which we find very exciting. There is huge potential to make this soft prosthesis at a very low price for low-income families affected by amputation”
The hand's design resembles an inflatable robot from the cartoon "Big Hero 6." The artificial hand is made of a soft, elastic material—the commercial elastomer EcoFlex. The prosthesis has five fingers, each covered with fiber segments, similar to the jointed bones of real fingers. They are connected to a three-dimensional "palm" shaped like a human hand.
Instead of controlling each finger with electric motors, as in most neuroprostheses, the researchers used a simple pneumatic system to precisely inflate the fingers and bend them into specific positions. This system, consisting of a small pump and valves, can be worn on a belt, which significantly reduces the weight of the prosthesis itself.
The team developed a computer model to link the desired finger position with the corresponding pressure the pump must apply to reach that position. The researchers also built a controller that directs the pneumatic system to inflate the fingers into positions mimicking five common grips, including pinching two and three fingers together and making a fist.
The pneumatic system receives signals from EMG—electromyography sensors that measure the electrical signals generated by motor neurons to control muscles. The sensors are placed in the opening of the prosthesis where it attaches to the user's limb. They can pick up signals from the residual limb—for example, when a person with an amputated limb imagines clenching a fist.
A special algorithm "decodes" the muscle signals and links them to common grip types. Engineers used this algorithm to program the controller of their pneumatic system.
When a person imagines holding a wine glass, the sensors pick up residual muscle signals, which the controller converts into corresponding pressures. The pump then applies this pressure to inflate each finger and achieve the desired grip for the user.
The researchers aimed to add tactile feedback—a feature not included in most commercial neuroprostheses. To do this, they attached a pressure sensor to each fingertip that generates an electrical signal proportional to the sensed pressure when touched or squeezed. Each sensor connects to a specific spot on the person's residual limb, so it can "feel" when, for example, they press with the thumb rather than the index finger.
To test the robotic hand, the researchers recruited two volunteers, each of whom had an upper limb amputated. After putting on the neuroprosthesis, they learned to use it by repeatedly contracting their hand muscles while imagining making five common grips.
After completing a 15-minute training session, they were asked to perform a series of standard tests to demonstrate hand strength and dexterity—stacking checkers, turning pages, writing with a pen, lifting heavy balls and picking up fragile objects (such as strawberries). They repeated these tests using the more rigid bionic hand and found that the inflatable prosthesis was just as good or even better.
One of the volunteers was able to intuitively use the soft prosthesis in daily activities—eating food (even crackers) and handling various objects (laptops, bottles, hammers and pliers). He was able to safely maneuver the prosthesis to shake someone's hand, touch a flower or pet a cat.
The most exciting exercise: the researchers blindfolded the volunteer and found that he could tell which finger they touched. He also managed to "feel" bottles of different sizes that fit into the prosthesis. The team views these experiments as a promising sign that patients with amputated limbs may be able to regain a form of sensation and real-time control with an inflatable hand.
