Recreating Software as Open Source
We have found a few examples of how we could use machine learning and pattern recognition to recognize weak muscle signals. CoApt is a great example and we have been using their evaluation unit to help Lorelei exercise her arm and it works really well!
We now want to replicate this software and build an Open Source version that we can share alongside everything else in this project.
Requirements:
We are still scoping out the requirements - Help us :)
We have gathered a lot of advice, which I have recorded here: https://sites.google.com/site/ourkidscandoanything/advice
We have met @Agamemnonc who would like to help us and has been working on this problem already. He has code available here: https://github.com/agamemnonc/pyEMG
Once we have the Bridge built (which is one of our other challenges) we will be able to work on this.
We may preprocess and calibrate pattern recognition software using a desktop PC and then load the pattern recognition 'keys' to the Rasberry Pi as a mechanism to offload any heavy lifting to a more powerful computer.