This two-week period saw the most progress of the research project thus far. As the researchers got more skilled at collecting and implementing EMG data, successful prototypes for a wide array of researcher projects started rolling in. All in all, this period produced six highly functional prototypes that will eventually be turned into final EMG projects.
As the speed of the project picked up, it became clear to that a way to organize and document researcher progress was necessary. Necessary for the researchers to share work, ideas, and code, necessary to show the project advisors and other parties interested in the progress of NTX, and necessary for the future of the project to train new researchers. Researcher videos, accompanied by a short description and other relevant information, have been organized under the "NTX Repository" page. I highly recommend that all readers take a look at that page and make their way through the videos (links have also been provided below in the bulleted section).
Instead of going through the sequential accomplishments of group as I usually do, I will talk about the implications of and our plans for the main technologies we are working on.
A while back we challenged the researchers to brainstorm some high-level EMG implementations with the eventual goal of making them user friendly, easily reproducible with a low cost, and highly impactful. We have been working on these project with the goal of having several completed prototypes with setup time less than 30 seconds to be ready for the Rochester Mini MakerFair in the middle of November.
EMG Controlled RC Car / EMGo Wheelchair
Team Software member, Niko Procopi, has made his primary deliverable over the last two weeks building and EMG controlled RC car. This car will serve two purposes: 1, a way to showcase EMG in an easily understandable way at events like Maker Fair and Imagine RIT, and 2, to serve as a precursor to our EMGo wheelchair; a controller box that will fit any standard electric wheelchair and allow the operator to control the device using facial muscles. As the videos document, we have accurate control of the device using two people (one to control forward and backward and one to control left and right) and serious progress toward making it controllable by a single user. This required two MyoWares to be connected to each of Niko's arms representing a major step to getting multiple methods of control from one muscle group.
Hillary Li from Team Electric has spearheaded a multi-function NeoPixel bracelet that changes color or design based off of the EMG patterns it detects. This is already a great tool to show visually how EMG works, but it has huge potential to be used for ASL as well. We plan eventually for this to turn orange when the sign for 'tiger' is detected, as well as trigger a bonus feature which researcher, Kenny Kim is working on. One of the most important early outcomes of this is the successful use of EMG gradients, as seen in the video. This will be hugely important for actions like acceleration in the EMGo Wheelchair.
EMG Morse Code
On October 12th researcher Spencer Logan created the first assistive communication device for the NTX project. The simple system detects the length of a muscle contraction to determine a 'dot' or a 'dash' and then prints a letter on the screen. While the accuracy was around 60-70%, this is a great first step and we will soon be ready to map this to a jaw clench.
Technical Project Lead, Colin Fausnaught, demonstrated successful control of a robotic arm using a MyoWare board. This will spawn into two projects, the Prosthetics Under $100, and the Advanced XAssist robotic arm. The former project will offer one mode of control, open and close, for an incredibly low cost and the second will allow for indivudual finger control. Both projects are in partnership with RIT's PORTAL Lab.
Advanced Electrodes & NTX Hardware
As these projects progressed it quickly became apparent that the MyoWare boards would not longer be sufficient for the advanced input. In addition to their low flexibility and reliability, they are clunky and are not easily setup by the end user. At the core of all of the above projects will need to be a new way of collecting and processing EMG data.
Researcher Reid Kovacs along with Colin are working on making fabric electrodes that can be woven directly into tight-fitting clothing. This ensures that the end user does not have to struggle with placing the electrodes, increases the number on nodes in a given area, and significantly lowers the cost. Early testing has shown promising results.
In addition to this, Team Electric, lead by Russell Cobb, have been working on building our own EMG / EEG boards and amps. This would allow us to further cut costs and build more efficient hardware directly for the need.
These weeks have been stressful and busy, but have shown great returns. Issues with ordering have been completely resolved from the hard work of Dr. Phillips, Debbie Monaghan, Mike Buffalin, and new addition to the team Jake Ellis. Without the hard work from all of these members we would not have made the great progress that we did this week.
This Period's Accomplishments & Events
Accurate 4 directional control of an RC car 10/7 (Link)
Open/close action for prosthetic arms using MyoWares 10/7 (Link)
First successful test of conductive fabric electrodes 10/7 (Link)
Jacob Ellis joins the team as Chief of Operations 10/9
EMG-based Morse Code test 10/12 (Link)
Researcher time and deliverables tracking sheet setup 10/12
Successful EMG control gradient collection and implementation 10/13 (Link)
NTX Repository built to showcase project and train future researchers 10/13 (Link)
Single user EMG controlled RC Car 10/13 (Link)
Advisor Caleb Ng Develops NTX logos 10/14
First Team BioMed Meeting 10/16 (Deliverables)
2nd Test of conductive fabric and new electrode design 10/17 (Link)
OpenBCI headset frame built 10/18