Finish Glove, Servo Box, RC Servo DAQ, editing program
This week I have finished the Sensor Glove and the Servo Box. The Box was all glued together, and one of each of the higher torque servos were screwed into the box. The Glove was finished. All of the sensors are in place, attached to their leads, and the connection wrapped in heat shrink to keep it all together. I am feeding the leads to the back of the hand inside the glove. I was worried about the leads moving and coming out each time someone puts on or takes off the glove. To help with that, I bought glove liners for someone to wear while using the glove. It is actually fairly comfortable.

This week, Chris received a new USB DAQ to beta-test for National Instruments. This DAQ has built in RC Servo control. This is a huge step for us, as a huge hurdle for this project was figuring out how send the servos data from my sensor inputs. This helps to take out some of the guess work.

Chris and I have played around with it, but it is still a work in progress. The input for the new DAQ does not work yet, but Chris figured that we could combine the DAQ boards we have, and use the older model for the input, and the new for the servo output.

The next stage is just taking all of the LabVIEW VIs I have and combining them into one cohesive VI.
Servo Box
I decided that it would be a good idea to construct a box that the servos and fingers could be mounted in, to give the project a more complete presentation. Ethan drew the box out in AutoCAD.
Sensor Glove
This week, I started to make the sensor input glove. I went to Home Depot and bought a pair of work gloves that I thought that would not hinder the playing of the pianist, but that I also thought I could connect my sensors on fairly easily. The pair that I found worked great. My friend Liz helped me to open up a fabric pocket on each finger tip, and insert my sensor. The first attempt was to have the lead come out of the pocket towards the wrist. But after putting one sensor in place, I realized that the leads would be moving each time a finger moved. The second idea I had was to have the lead come out of the finger tip and wrap around to the top of the hand. This idea seemed to work much better.
Check In
Ethan and I met with Chris today, just to discuss our project and to outline our plan for the next few days.

What we have accomplished so far:
Read force data from striking the key with LABView.
Prototype some finger designs - two of the same shape but with different cushioning at the tip.

What our next steps are:
Mount the servo arms onto the fingers.
Build a housing for the servos.
Finger Design Part Deux
After my meeting with John, I decided that a redesign had to take place. Before, and from the video Ethan and I took, we decided that when the finger repeats a not on the piano, that there are two joints involved - the wrist, the Radiocarpal Joint, and the joint at the knuckle of your index finger, the Proximal Interphalangeal Joint. (Click here for more)

From the information that John gave me, when properly playing, and just repeating one note, the only joint involved is the Radiocarpal. He went on to further explain that you would want to keep your hand and finger rigid, in a J or hook shape.

I took this new information and using the laser cutter, which rocks if you have never used one, to prototype two fingers. I used the hand model that Kevin provided, and tried to take as accurate measurements as I could. I was using a ruler, so the dimensions are only accurate to 0.01 cm. If further fingers are made, I will use a caliper with more divisions.

Taking these two fingers, I tried to see what kind of data I could read in LABView. I found that the data was similar to when I struck the piano action we are using for testing, but I noticed that the sharp edges of the finger were making small indentations in the FSR. This worried me as they could potentially damage the FSR, but also change the sensor profile. If I were to strike the sensor in the same place as a dent in the future, any change that the dent made in the structure of the sensor could potentially change the readings. To deal with this, I fixes a small plastic or rubber foot on the end of one finger. It is similar to those found on the bottom of some cutting boards to prevent sliding. On the bottom of the other, I took a small piece of rubber from an ankle brace insert, cut a small circle, and hot glued it to the end of the finger. This hopefully will help protect the sensors, but also allow our fingers to act more human like - some amount of force may be dissipated, resembling the flesh on our fingers.
An Interesting Development...
Today met with John Arcaro who gave me some more insight into playing the piano. He is an Assistant Professor at the Berklee College of Music in Boston, and he explained that if I want a motion that will just play one key, repeatedly, then the motion really comes from the wrist. This helps to simplify the project a bit. I can now construct a single solid static finger, and only have to worry about one servo actuating the writs joint. This is a good stepping stone for later, when more complexity is to be added to each joint.
LABView
For the last few days, I have been trying to get the LABView program working. I had to install the drivers for the DAQ on the computer, and then started to build the program up. I must not have been doing it write, or could not figure out what the icons were supposed to be, as I had errors even when I had it setup just like the examples in the included documents. Paul gave it a try, and he was not sure what the error was either! Anyway, I had sent an e-mail to Ryan Schoonmaker, who with I took Chris' Data Acquisition class. I asked him if he still had the program from our project in that class, the Sense-O-Sock, as I seemed to have deleted it or it was lost among my many files. I knew that we had the DAQ working in that case, and Ryan had written this awesome bit of code that dealt with graphing the force felt by the FSRs we used then.

I had to edit the program a little, but I am now getting the data I want from the system. When the piano key is struck, I am able to determine the amount of force used. Now I have to start working on the feedback part of the system, but we have to have the finger/servo setup completed first, I think.
Design Ideas - Results
Ethan and I met with Chris and Eric, who is a graduate student who is running Robotics Academy. We presented our ideas, the designs and constraints, and here are the results from that meeting:

- It was recommended NOT to use the NITINOL wire. Although I found a company that sold the wire, as well as compression springs, Chris did not feel that they have a fast enough response time for this particular application.
- The finger design that everyone seemed to think would function the best was a sort of linkage. Check out the Pictures and Drawings Section, under the Multimedia tab, for a cleaner drawing (no red pen!).
- The software interface will be LABView. This is great because I have had a lot of exposure with this, from taking Chris' class last semester, and that he is the project advisor! We will also use a USB DAQ from National Instruments (NI USB DAQ) to collect the data from the sensors.
- We will use FSRs (Force Sensing Resistors) to determine how much force is applied to the piano keys, and Bend/Flex Sensors to also help determine this value.
- Servo Motors will be used to actuate the finger.

Ethan and I will now go and figure out what exactly we need to order.
Design Ideas
Ethan and I are meet today to discuss finger designs and constraints. We are meeting with Chris tomorrow and I want to present him with at least two finger designs, and Ethan is working on a list of constraints.

For the designs, I have come up with three:
- A linkage driven by servos
- A finger that will flex and extend by NITINOL wire, or muscle wire
- A finger actuated by string and servos

Some of the constraints are:
- For a simple, repeating motion, we only need 1 Degree of freedom
- Materials - we want to closely mimic the finger in size and weight
- Cost of materials - for the course part of this project, ME 43, the Mechanical Engineering Department allows each student to spend ~$50. Being that I am the only senior, this project would be extremely hard to attempt without Chris's help.
Research Continues
Ethan and I had a short meeting today to exchange files and documents we found, as well as discuss what we want to present to Chris. We decided that it would be a good idea to have at least three solid finger designs, as well as input and contraints,
Research
First, I want to inform you that I will be working with Ethan Heller, a junior mechanical engineering student, on this project. Hopefully he will be able to ad to this journal.

With the kind of task, the first thing that really has to be done is research. Ethan and I have spent the last few days searching journals, in and off the Internet, for past projects and Robotic Hand experiments. We found that there is a great deal of information out there - meaning that there is much to sift through to find articles that pertain to our project and research.

We have found that a lot of the robotics hands that are out there use a system of cables and pulleys to control the finger movement. This is a good system for grasping, because once the cables are tensed, the fingers will close, grasping an object. Keeping tension in the the cables allows the hand to continue to grasp the object. Our motion is slightly different. Our goal is to design a finger that will play the piano, and allow for variability in the speed in which the finger plays the key.

After looking at all of the previous research, Ethan and I should be able to come up with a few different ways to approach this problem.
So it Begins
With the start of my senior year, there is the added bonus of being able to participate in and construct our very own design project. This blog will help to catalog the experience, and the project.

A little background information may serve well here:

I am Andy Margules, a senior mechanical engineering student, and part of the Tufts University Robotics Academy. For those who don't know, the Robotics Academy is a subgroup of the Tufts Engineering Department. For more information on the Robotics Academy, please go to http://ase.tufts.edu/roboticsacademy/

I was presented with this project by Professor Chris Rogers. Here is the project as I understand it. Steinway & Sons, one of the most prestigious piano manufacturers in the United States, hand makes all of the parts of their pianos. Due to this, there is some variability in the size and shape of the piano actions. A piano action is the mechanism composed of the piano key and the hammer which strikes the wire creating the sound of the note.

To provide a level of quality control, Steinway tests their piano actions. I have been informed that their testing method involves a metal bar, probably made of Aluminum, attached to a servo that repeatably hits the key.

I have been challenged with designing a more humanoid device to test the piano actions, as well as allow for Steinway to test each key with different amounts of force.

Should be a fun project.