I will make two anatomically correct arms from the shoulder to the fingertips which will be motorized. I have been doing research on this project and am interested in insuring that the arm itself is constructed with proper proportions and that the mechanism which will move the arms will respond in as natural and realistic a manner as possible. These arms will be displayed as disembodied and spaced appropriate distances apart. A kinect will be responsible for triggering movement.
Other, much more intelligent people have devised interesting mechanisms for soft-jointed robotic movements.
Such as this group: Washington University where they reproduced highly dexterous movements….
And then there is this thing:
This project touches on so many of the aspects of the arm that I would like to touch upon: specifically, the smooth motion of the shoulder.
Please please go to the link above and check this out.
The final project I will be linking to is this:
They all seem to be doing what I thought would be right – using a flexible material to execute the movement in imitation of the muscles and tendons of biological joints.
Also: there is a decent instructables tutorial on biomechanics:
I want to have people imitate flight (like people do), similar to this project below:
This motion is meant to be a streamlined one from the natural pose that people take with the Kinect (the arms up “freeze” stance) and use that as a natural segue to the motion of the interaction: It is a form of waving at machines. It is also futile, a human waving their arms can no more fly than these two arms will ever be real arms. When people pretend to fly in front of the arms, the arms will mimic their motion in real time.
I will, for this class, only be motorizing the shoulder joint and the elbow. The wrist and fingers will be on joints which will respond to gravity and the momentum of the elbow and shoulder.
I want to construct the arms in a beautiful way, one which does not aim for realistic rendering of the surface, but rather which highlights its form. I find topographical maps to be amazingly informative but also beautiful.
This image has variants of surface and color. What I will be doing is rendering the arms in paper alone. My past work dealt a lot with stacked paper, forming bas-reliefs:
I intend to reinvent my older working style into this arm and use the typographical map as my inspiration. That being said, I need to understand the form of the arm as much as possible. Our awesome librarian linked me to several web sites devoted to 3D visions of the human body.
Body Parts 3D is one which I found extremely beneficial.
Having problems articulating these images into a rounded cut-able form I discovered an autocad site which creates a 3D model based on photographs you give it.
This was my first attempt at a capture: It rendered the space around my arm.
My second attempt with 123D Catch rendered me a 2 palmed hand:
Though the image is super cool, it does not serve me well for this project. I ended up dropping 123D Catch (though will return to it) for already-made high quality pre-rendered arms. I have a right arm and a left arm and it appears to be anatomically accurate. (the screen shots are in Rhino).
I need the arm to be flexible in the joints. So I must sever the mesh object and reform it with hinges of paper. There are an amazing amount of resources for paper folding, hinging, cutting etc. I am not familiar with Rhino, but below I am removing faces from the mesh- tracing where the fingers fold.
Though this does not sever the mesh, it is how I intend to do so. Once it is severed, I will make a joint based on how thick the paper cuts will be (the housing for the joints etc will be inside the paper). If the housing needed is small, then the paper thickness will be greater, and my hinges must change in accordance.
This eventually will go back into 123D- this time 123DMake. This service provides you with a cut list for laser or CnC or will send the files to a 3D printer. In other words, 123D Make is a free program for easily translating your 3D files into translatable formats for your digital fabrication needs.
In the interim of sourcing the external aspects of the arm, I began researching and playing with the motorized aspects:
The UNIVERSAL JOINT! (these are so precious).
These universal joints are an almost replica of the human shoulder joint.
These little guys are so fluid and to have a motor stuck on them to create movement seems like a waste. (i.e. limit the universal joint to servo movements)
Servo users has attempted to solve some of the restrictions of the servo by giving them a pan and tilt housing. This joins two servo motors for different motions for one object. The one below are mini servos and the housing itself is really crappy.
The one on the bottom (left in photo) is responsible for the “pan” and the one inside the housing is responsible for “tilt”. I don’t think I will use this specific brand.
However, in thinking how to join these two (the universal joint and the pan/tilt servo combo) I looked to the human body itself. Tendon Muscle relations are complicated as are our human joints themselves (there is liquid and and and) which the mechanized version will not be able to recreate. The ligament helps to transform the movement from the muscle to the part of the (arm in this case) which is to be moved.
So, I strapped the universal joint in a vice:
(just so you know, the length of the metal is an accurate length for the upper arm)…That green thing- It is a replica tendon (i.e. rubberband).
Here is a video of only one “tendon” and one motion of the pan and tilt servo (manually manipulated). Though it is wild and wonky, the movement of the steel rod is much more fluid than it would be directly mounted on the servo.
My next steps are to recreate this but breadboard the servos and research where tendons are in the arm and try to make a motor connect properly to them.