In the past two weeks, a lot of progress was made towards our final product. During the Thanksgiving break I worked with Kayla to build the classifier that would be used to interpret the muscle input signals and classify which movement it belonged to. We integrated together the feature extraction that Kayla had been working on along with the SVM model that I had been working on. Afterwards, we integrated our classifier with the game that Alex had been developing for our third integration test, to see how all of the pieces of our project fit together. In doing this, I realized that one of the electrode amplifiers in the system had stopped functioning properly (it’s travelled around the country quite a bit in the last couple of weeks, which could be a reason why) so I had to remove it from the system and solder together one of the spare electrodes we had purchased at the beginning of the semester. (shoutout to the past versions of us for thinking ahead in the event that this should happen). 

During this past week, we worked to improve the accuracy of our system to meet our target metric. To improve the classifier, I added a priori distribution of ground truth data, which Byron had suggested that we make during our first demo. It’s come in quite handy in making our classifier more accurate. This would be a great place for a picture but all of this basically happened in code so if you’re interested then check out our github repo. We also did another in class demo, during which we got some feedback on how to proceed with our final presentation. In the past couple of days we have been working on our final presentation, and during this weekend we will finish up final touches on our project to complete our presentation. 

In this coming week, we will as a team focus on delivering a cohesive presentation, making our demo video, and writing our final report.