-
-
Setup
Inspiration
Escaping reality via video games is something a lot of us enjoy. Unfortunately, some of our bodies don't cooperate and trap us within. We at Integrating Factor wanted to break free. We wanted to create an interface that is consumer friendly and takes a proactive approach to detecting, identifying, preventing, and in the future, curing ailments that impact the brain and body.
What it does
Simply put, PlayMG lets you interact with a computer without even touching it, by just moving your body. It uses EMG sensors to detect changes in muscle orientation, along with a set of kinematic sensors to identify specific movements. However, that's not all that its capable of. In order to solve the problem of detecting and preventing degenerative diseases, we need identifying data, lots of it. By framing the collection mechanism as a control interface, the user-base can scale indefinitely. Anyone can interface with a computer using PlayMG, whether they are fully able bodied, or suffer from any number of ailments. Usage of PlayMG will create individualized electromyographic profiles and detect changes in motor function over time. This data has a number of applications. Whether its detecting improper ergonomics at your desk and common ailments like carpal tunnel syndrome, or discovering and adequately quantifying a decline in muscle function which may be caused due to any number of diseases, like ALS, Arthritis, Parkinson's, etc. This data can also be stripped of identifying info, only including critical demographics like age, height, weight, plus any pertinent ailments, and then be used as a data-bank available to researchers and medical professionals. This data would prove invaluable in development of new diagnostic and therapeutic techniques to provide relief to those afflicted.
How we built it
Using the BioAmp EXG Pills, we used a sophisticated arrangement of EMG electrodes in the forearm region to study the signals generated by different muscles. We conducted a comprehensive set of experiments in order to clean up the data and attempted to identify different movements through envelope analysis in the signal collected. We have included some of the collected plots in the github submission. In addition to the EMG data that we collected, we included a couple of flex sensors installed on a glove to narrow down the scope of movements and fine tune our signal identification.
Challenges we ran into
The first major hurdle we faced was cleaning up the signal. Since none of our team members possess a neuroscience or biology background, we had to conduct a significant amount of research in order to figure out how EMG even works. Once we got around to collecting the signal emitted by the muscles, we found that it was very polluted. This was due to a goofy mistake on our end, and resulted from a mislabelled input. However, before realizing this, we were able to fine tune our algorithm to the point where we were able to detect muscle twitches with a 97% accuracy. Post labelling fix, the data we collected was incredibly crisp and responsive. Unfortunately, we were unable to consistently identify specific movements through EMG data alone, as our signal analysis was not sophisticated enough. This brings us to why we used flex sensors. They were intended to help us narrow down the scope of movements associated with inflections detected in the signal. Unfortunately, the flex sensors failed during testing and did not produce the full range of resistance values forcing us to discard their use in our final test. With the use of a better movement detection mechanism and further study of signal and biology, we are confident that we can perform complicated operations through the PlayGM interface.
Accomplishments that we're proud of and what we learned
We were able to collect remarkable data and created an actual brain computer interface that we can use to interact with our computers. Even though we didn't meet our goal of wanting to play Doom with our interface, we were able to get over about 3 obstacles in Geometry Dash. The learning curve was steep, and we could not have accomplished anything without the support of our teammate Fawwaz Hameed, who did almost all of the programming on his own, and did a remarkable job at it. This allowed the rest of our team to work on research and the hardware end of things, and gave us room to experiment and learn from our failures.
What's next for PlayMG
If given the opportunity, we want to further develop the interface by employing more sophisticated signal analysis techniques and find a teammate with a specialization in neuroscience. We would like to spend more time on hardware development in order to have a fitting complement for what our software can achieve. Lastly, we would like to start with creating individual profiles for our team, and branch out to people at the university and beyond, with the final goal of commercializing the product and achieving our vision.
Log in or sign up for Devpost to join the conversation.