Browsed by
Author: kgv

Product Pitch

Product Pitch

Comments 0 Comment

It’s estimated that 42 million people in the U.S. suffer from some form of movement disorder [1]. Whether this requires rehabilitation of the nervous system or a form of prosthetic, a common challenge that follows is proper muscle rehabilitation. We created the game MyoRun that combats deterioration of muscles and improves muscle control through continuous play. This game is an electromyography (EMG) controlled endless-runner style video game. With innovative EMG technology and an easy-to-use interface, this game is interactive, fun,…

Read More Read More

Kayla’s Status Update for December 5th

Kayla’s Status Update for December 5th

Comments 0 Comment

This week, and including last week, I worked on narrowing down the feature extraction and classification pipeline to be used for classifying EMG signals in our project. After discussing with Tarana to understand the expected data input to the SVM classifier, I created a 1×15 matrix for the feature vector to which included the max value for each electrode, five binary signals about whether or not a particular electrode fired during the recorded signal, and the peak frequency from the…

Read More Read More

Team Status Update for November 21st

Team Status Update for November 21st

Comments 0 Comment

This week as a team we are putting pieces together before our third integration test. Early next week we plan to do this final integration test in order to be set to work on the rest of our deliverables for the final submission and demo of this project. We are all moving this/next week because of the holiday so that has been taking up some of our time. Major things that happened this week on the signal side, feature extraction…

Read More Read More

Kayla’s Status Update for November 21st

Kayla’s Status Update for November 21st

Comments 0 Comment

This week I have been continuing work on the machine learning aspect of the signal processing. I obtained my first accuracy result from splitting our 42 trials of data between training, validation, and testing data with an accuracy of 35% using a Long Short-Term Memory (LSTM) algorithm classifier. This definitely does not meet the accuracy we are aiming for so there is still work to be done in the adjustment of hyperparameters and the input feature vector. In my further…

Read More Read More

Kayla’s Status Update for November 14th

Kayla’s Status Update for November 14th

Comments 0 Comment

This week I dove into applying different data dimensionality reduction and classification methods to our data. First, with the primary data that we have collected, I created a visualization of the peaks of the recorded signal data in order to see if there might be any obvious trends in the correlation between movements and the peak of their signals. This graph shows the max value for the voltage recorded by each electrode for each sample of each movement from the…

Read More Read More

Team Status Update for November 14th

Team Status Update for November 14th

Comments 0 Comment

This week our team presented a demo of our project to our TA and professor. The main components of the project that we demonstrated were the relatively complete subsystems of the hardware EMG data collection and the software system visuals of the game. The final subsystem of the project that is in progress is the signal classifier. We got feedback from Dr. Yu that we can generate simulated signal data in order to verify the functionality of our classifier without…

Read More Read More

Kayla’s Status Report for November 7

Kayla’s Status Report for November 7

Comments 0 Comment

This week for capstone I spent the majority of my time organizing the classification algorithm approach to our EMG data. A major decision factor going into this classification of the data depends on how the data is recorded. I found that if the data is recorded in a continuous stream of repeated actions, the best approach is to use wavelet transform on the data in order to identify a basis function and the scaled and shifted versions of it. This…

Read More Read More

Kayla’s Status Update for October 31st

Kayla’s Status Update for October 31st

Comments 0 Comment

This week the data collection process was bolstered and thanks to Tarana’s hard work in coordinating the hardware system of the EMG, we were able to collect data recorded from more than a single electrode. Last week’s milestone was collecting data from a single electrode and visualizing the signal from different recorded movements over time. This week, toward the end of the week we got data collected for all the movement collected from five electrodes on the forearm. This upcoming…

Read More Read More

Kayla’s Status Update for October 17

Kayla’s Status Update for October 17

Comments 0 Comment

This week I presented for our team’s design review presentation. A question that came up was what is the difference between game delay and total delay. These names sound quite similar which can be confusing, but the metric for game delay helps to isolate time required for the game to render and react to the mock muscle simulator, while total delay is measured when using actual muscle signals which must go through the classification algorithm. Identifying these metrics and risk…

Read More Read More

Kayla’s Status Update for October 10

Kayla’s Status Update for October 10

Comments 0 Comment

This week I spent time furthering specifics for the data collection experimental procedure of our project. Additionally, I have been exploring common data processing techniques for classifying EMG data, specifically useful feature extraction techniques such as Wavelet Transform. I will also be presenting during the Design Review Presentation for our team next week, so I have been working on the slides and preparing the presentation. Our project is currently on track and the next step is to collect the EMG…

Read More Read More