The lab's first underactuated prosthetic hand

A hand has many more joints than can be motorized in an affordable prosthesis. This project solved the problem with underactuation: a single actuator pulls a cord that, through pulleys, shares the motion among the 14 finger joints. The prototype achieved stable, adaptive grasps of objects of different shapes with four fingers, and showed that the thumb needs separate treatment.

Context

Commercial myoelectric prostheses use several motors and cost tens of thousands of dollars. Hook prostheses are cheap and functional but offer a single type of grasp. An underactuated mechanism seeks a middle point: few actuators, and mechanics that distribute motion according to what the object allows.

Mechanism

Each finger has three phalanges with pulleys at the joints and torsion springs that return it to the open position. The actuation cord runs over the pulleys of each joint.

In the palm, a system of intermediate pulleys splits the tension among the fingers.

When a phalanx touches the object and stops, the cord keeps running and the others keep closing. The result is a grasp that conforms to the object's shape without sensing or control at each finger.

The thumb has an additional degree of freedom, the carpometacarpal joint, which the user sets by hand in one of three positions depending on the type of grasp.

Prototype

Parts were 3D printed and the tendons were threaded through the channels in the palm.

Proposed control

The design includes a complete loop. The electromyography signal sets the force command, a strain gauge measures cord tension, and a vibration motor tells the user how much force is being exerted. The project was published under an open hardware license.

Results

The prototype performed stable grasps of objects of different geometries, with the phalanges adapting to each one.

Testing revealed a limit of the design. With the cord path planned for all five fingers, the input force needed to fully close the thumb and the other four was too high. The choice was to actuate only the four fingers from the intermediate pulley in the palm and leave the thumb unactuated, as a passive support. Even so the thumb proved indispensable: it is what supports the object and stabilizes the grasp. Including it also forced one pulley channel to be tilted, which increased the changes of direction of the cord and the friction.

What is missing

Friction in pulleys and channels consumes a good part of the actuator force. The thumb is not active. The myoelectric control loop with feedback was designed, but its validation with a user is not part of this work.

How it fits in Robiolab

This is the origin of the lab's prosthetic hand line. The three following projects tackled, in order, the limits this one identified: the mechanics of the hand, with a floating body that reduces friction and an actuated thumb; the force sensor and its controller; and the integration with myoelectric control and a user test. The central idea, that adaptation to the object is solved by the mechanism, is an example of the mechanical intelligence the group looks for in its bioinspired designs.