Robotic Hand Project

I have been interested in bio mechanics for a long time and I wanted to design a nitinol actuated robotic hand. I wanted to focus on a hand design that emulates how real hands are actuated. I am familiar with finger tendons from my sport which is rock climbing, and it was really interesting to learn more about how they work together. 

I did a deep dive on hand anatomy specifically looking at flexor and extensor tendons to understand how they actuate and are positioned on the hand.  I knew that my first prototype would be nowhere near as complex as a human hand, so I was mainly focused on the fundamentals of the "cable driven" system. 

I designed a hand with all of the actuation in the finger joints. First I planned out the CAD of a single finger of the hand with three ligaments per joint (see left image). These ligaments would need to be printed seperately and bolted on. The amout of fasteners that would be needed to assemble an entire hand this was was a bit overwhelming. 

This issue made me extremely interested in streamlining my manufacturing process using Cura's "Generate Interlocking Structures" feature. This feature would allow me to weave TPU ligaments into rigid PLA bones of this hand, which would act as a flexible fastener to hold the bones together while also still allowing actuation. I revised my design to house print in place ligaments, with an example on the right showing how these ligaments would interlock. 

I printed my first prototype and it was less flexible than I expected. I concluded that this was because I underestimated the stiffness of the TPU 95A I was using. A bigger issue with this design was that the ligaments did not interlock deeply enough into the bones to make a strong connections. The small pieces of TPU that were woven into the rest of the hand broke quickly when the fingers were bent. 

I saw a really cool Youtube video titled "Alternative to bearings for tiny robots" that inspired me to approach the joints problem differently. The video was created to bring awareness to rolling contact joint as a concept, and I thought they would be super applicable to my design since biological joints are often modeled as rolling contact joints for their low friction. The joints I saw online were held together with cable of some kind, but I wanted to maintain my streamlined manufacturing process. I designed a multi-material print that used the TPU as cable for the joint. I used the interlocking structure feature again for this prototype. 

My first prototype worked, but the interlocking structure which extended along the entire cable interface which restricted it's range of motion. The cables were connected to the "bone" everywhere they came in contact with it. This caused the rolling I was trying to achieve to be more like a pivot.

I modified the cables to only be connected to the bone by tabs at the end of the cables, which allowed the joint to achieve significantly more rotation. 

Real joints use non-radial surfaces to achieve non-linear motion. My next objective is to create joint geometries that achieve motion similar to real finger joints. Above is my first iteration of a non linear joint, which rotates smoothly, but actuates too much in some cases due to the lack of tension in my "cables". I plan to address this by adding additional tabs to connect these cables to the bones in strategic locations to contrain the movement. 

This is an ongoing project, and I am really excited to continue using it as a way to learn more about bio inspired mechanisms and movements.