We both are very health-conscious people who love programming -- which often entails sitting at a computer for hours on end. A very common problem among people like us is ergonomics, mainly carpel tunnel and back pain. Some software engineering companies hire in house ergonomists in order to support their employees. We thought it necessary to build a framework which could help everyday developers improve their ergonomics and maintain proper body positioning to stay healthy through their long programming sessions, almost like having an on hand ergonomist at all times.
What it does
Using Thalmic Laboratory's Myo to track hand and forearm positions we warn the programmer when they are in a position that can cause damage to their carpal tunnel. Using OpenCV and a webcam we ensure that the programmer knows when their posture can lead to high back stress. Furthermore, we provide reminders relating to health throughout the hacking/computer session.
How I built it
The Myo's sensors provide an abundant amount of information about the position of the wearer's hands. Using this information, we train Neural Networks so that our service is tailored towards each user and can best identify their specific ideal arm/hand positions. We used another Neural Net to again tailor our service's posture identification to every user. Most of the program is written in NodeJs but we have a number of scripts written in Python that interact with the Node app during runtime.
Challenges I ran into
Neither one of us had any hands on experience with working with big sets of data and training neural networks, so this was a huge challenge. More difficult was figuring out that more information isn't necessarily better - we had to ignore more than half of the available information from the Myos and plenty of pixel data from the webcam. Furthermore, we had some trouble connecting the Python portions of our project with the Node.JS portions to form one cohesive experience.
Accomplishments that I'm proud of
We worked really well together to create a project that we both found important and interesting, while learning entirely new concepts along the way. By far we are most proud of managing to train algorithms to help prevent problems we are all too aware exist (if only we had Ergomyonic during this hackathon, cause we're definitely feeling the effects of 36 hours of programming).
What's next for Ergomyonic
The sky's the limit baby.