Finger Mechanism
In this post I’ll explain the finger mechanism functionality for the compact bionic hand.
The fingers are tendon driven by small dc motors. Each finger has its own motor which loops an artificial tendon around a small custom pulley. The material I used for the tendons was Dyneema synthetic rope - this is strong and has minimal stretch. Material like Nylon fishing line can stretch over time under high loads which would lead to loose and sloppy finger actuation.
As seen in the cross sectional image of the index finger below, the red line represents the flexion tendon. As the motor rotates the pulley, the tendon wraps around it, effectively shortening its length and tensioning it. As this happens the finger curls inwards/closes, or flexes. As long as the motor is powered and holding this position the finger will remain flexed and provide some holding force.
The green line in the image above represents the extension tendon. I decided to simply use a piece of elastic cord for this. Once the motor releases tension on the flexion tendon the elastic cord (which has been stretched) will provide a restoring force to extend the finger and make it straight again.
It’s important that the channels for these tendons have clear unobstructed pathways in both the extended and flexed positions. This prevents the tendons from being pinched between moving parts.
The image to the right shows the tendon pathways at the tip of a finger, showing curved open transitions at the pathway interfaces between finger segments
The fingers were joined together with small steel pins which were press fit into place. I purchased these pins from Accu.co which conveniently sells them in various diameters and lengths. I also incorporated nylon washers in each joint to reduce friction. The fingers were SLS 3D powder printed and I thought the friction of prints rubbing on each other wouldn’t be ideal. You may wish to view the joints in more detail in the CAD files
I designed the base part of the finger (or the knuckle part) to have a profile than can slide into the palm piece. This allowed me assemble each fingers with the tendons in place and simply slide them into the palm. I later used a back retaining plate to hold the fingers in place.
The motors I used to drive the fingers were micro metal gear motors from Pololu. I believe these types of motors are generically called N20 motors. It was important that I could drive the motors to specific positions and be able to detect/measure what position they were at. For this reason an encoder was needed. I wanted to use an absolute encoder as this would retain position information in the event of a power cycle. The encoders I used were micro rotary potentiometers manufactured by Bourns.
Normally one would usually use a hall effect or optical encoder coupled to a rear output shaft of these motors (which Pololu also sells). However, the drawback of this was that if the device lost power the motor/finger location data would be lost and a calibration cycle would have to be run again. This obviously wouldn’t be ideal for a prosthetic device but in hindsight would have probably been fine for an early prototype.
As seen in the image to the right we have the motor with its inbuilt gearbox, the 3D printed pulley (purple) and the micro rotary potentiometer (transparent).
One negative of using the micro rotary potentiometers was that the motor shafts were 3mm D-shafts and I could only find potentiometers with a 4mm D opening.
As seen in the image to the right the motor shaft is slightly undersized in the potentiometer opening. I added some protruding material to the 3D printed pulley to fill this area. This somewhat worked on the SLS powder prints I made, but because the walls are so thin (0.5mm) this would probably not work on a regular FDM print.
Furthermore, in reality its actually not that big of a deal if the motor shaft doesn’t fill the entire opening space. This is because the motor body as well as the potentiometer are held secure in the palm and the motor shaft rotates concentrically relative to the potentiometer, so in theory the shaft shouldn’t be wobbling around inside the opening. Also, the flats make contact so the pot should react responsively to motor movement (i.e. no lag between motor movement and positional readings when direction changes)
The image below shows the hand with the palm made transparent. We can see each motor subassembly in place. I had some channels through the upper part of the palm to allow the flexion tendon to pass from the fingers to each respective pulley. This isn’t clear in the image but can be seen in the CAD files.
One last thing to note was that I had access to a 3D printer which could use flexible/soft materials. The black finger tips were made from a squishy rubber like material for enhanced compliance and grip.
