What Am I Doing?
Build a hand prosthesis … from scratch
MORE DETAILS (not much)
The assignment was pretty broad; try developing something that could be of benefit. There’s not much more jargon to add (if any) or any more context associated. Pretty straightforward. My first thought was a hand prosthesis. (I’ll get into why I choose building a hand prosthesis)
What was I thinking?
I want to make hand prosthetics.
(little do i know … )
MORE DETAILS (context)
Not practical, to say the least, given the history of hand prosthesis development. However, I thought the choice struck a good balance between my ego (which at the time was … high) and … well, the choice was more to satisfy my ego.
After a bit of reflection (well after the assignment), I think I personally viewed the assignment as “how well can you design and make?”. The obvious difficulty and lack of adequate solutions (personally speaking) were the primary reasons for choosing a hand prosthesis build, as well as a small hope for a breakthrough (let me emphasize … small).
What I made?
A finger. Just one.
But it is the most detailed finger I’ve ever designed.
MORE DETAILS (WARNING: lots of text)
I call it “The Finger” because it’s one of the more technical models I’ve created. Let’s get into the details.
In the model, there is a single sketch that is called the “master” sketch. Much like a parent object in programming, changes to the “master” sketch have a cascading effect on its child elements. Using cascading effects, the “master” sketch changes the width and length of the distal, middle, and proximal phalanx (all three of which are part of the model). As you might suspect (for those who are familiar with CAD model development), sketches are not typically used to control the — entire — model. Initially, what I had in mind was a GUI to quickly change the different phalanges’ length and width. For SolidWorks, there is no such thing (as far as I could tell). The next best solution I could come open with was an open sketch. Change the open sketch dimensions, click save, and expect the entire model to update to the new dimensions. To my surprise, I worked out great, for the most part (a couple of bugs, i’ll get into them).
Forming the actual model (using extrusion, cuts, etc.) was pretty much standard, nothing too unusual. Although, I did use the “copy part” tool, for some particular reasons. It’s a bit of an advanced fix to some issues I ran into. However, I don’t want to concentrate on that. I will shift my focus to — “bugs”.
BUGS
Using the open sketch (to automatically adjust the different phalanges’ widths and lengths) has its issues. I ran into three major bugs: disappearing geometric constraints, colliding geometric constraints, and inversions on the geometric constraints.
Disappearing geometric constraints came solely from using the extrusions themselves as references. Adjusting an extrusion’s dimensions redraw the lines and vertices associated with extrusion, and delete all prior geometric constraints as well.
Moving on to collisions, the “collisions” occur when geometric constraints do not allow each other to move within their entire range of motion. After many iterations, the fix was to be selective and scarce with the geometric constraints used in sketches. Unnecessary geometric constraints were the main cause of issues when it came to smoothly resizing the model. The “collisions” stopped after following that guiding principle. Pat on the back.
There is still one more bug that is just too complicated for me to tackle (for now): inversions. I’ll give an example, to clarify. Let’s apply the geometric constraint called “collinearity” to two simple lines. The constraint can be satisfied in two arrangements: as it lands (done by the CAD software automatically) or by rotating one line by 180 degrees. It is difficult to predict when an inversion will occur, given that the programming behind transitions from one state of dimensions to another is not visible to the user. Once inversions do occur, they are not reversible. You must specifically undo the changes. Changing values back to the original does not guarantee a return to the prior. The issue of inversion mainly stems from shapes that expand beyond their intended region or shrink too much to the point where geometric constraints cannot be met. Ultimately, I found a range of values that did not cause inversions and stuck to those values (which were acceptable, given the typical dimensions of real phalanges).
There are a lot of details in describing bugs that need multiple visual references, to provide a full picture. So, i’ll stop here.
Looking back, I am glad I reached a point closer to the frontier of bugs, where they tend to be more elusive.
One last detail to explain. After completing the model, I resurfaced the finger to have the same texture as a real finger. The operation was pretty simple and straightforward. However, at the same time, the operation was a one-time fix. Working with organic models presents their own set of issues. The two main issues were resizing and fixating. The scanned finger model was difficult to automatically resize as the “master” sketch changes (in other words, not accomplished). Superimposing the scanned finger on the modeled finger with a fixed orientation was not simple.
Did I help?
No.
I tried to make a useful hand prosthesis.
I got stuck at the finger.
MORE DETAILS (explanations)
I was completely focused on two things: building an easy customization method and increasing the force efficiency. Before I knew it, most of my time was spent on a single finger. Luckily, the customization aspect of the finger model made it easy to create the different bone length ratios associated with each digit. So, effectively, I made finger-s.
I was a bit bummed I could not make something immediately useful, but hopefully the model was easy enough to be used as an add-on for other projects. Also, I found the model to be technically impressive, personally speaking. The model was tailored to be printed as a working joint and the customizability was acceptable.
Overall, I feel good completing such work. Even to this day, looking back at the model’s development, I find interesting design choices that seem ingenious, or something I didn’t think I had the creativity to do. I feel confident I can work on top of this work and ultimately make a grand prototype later on. I already started the next step (adding a palm). See the ECPI tab on the portfolio page for more information
(last words) This project provides insight on the bumps and obstacles in the transition from biological design to robotics, which is the ultimate goal. In other words (best posed as a question): how do I replicate biological designs, as a fabricator?
Big thanks to the Dr. Peter Pidcoe’s Lab at the VCU Department of Physical Therapy for being tolerant (it was my 1st internship) and for the support.