Myo Imaginary Keyboard Training Data Generator

This is a CLI program to create data correlating the raw electromyogram(EMG) data from the Myo armband to keyboard input. By feeding the data to a machine learning algorithm, such as an artificial neural network(ANN) or support vector machine(SVM), one could create a model to predict keystrokes based on EMG data, and be able to essentially type on any surface- an imaginary keyboard of sorts, limited only by the driver's inability to get raw data from two myos simultaneously.

Unfortunately, the length of time required for the models to converge without GPU acceleration is too long for me to complete debugging by the close of the hackathon, and the training data generator is being presented instead.

To run the program, execute TrainingDataGenerator.exe in the terminal, and experiment with keypresses- they will be represented on the side of the EMG data in ASCII number format. The output is either displayed on screen for demonstration/debugging, or piped to a logistic regression algorithm, such as an ANN or SVM.

Built With

• myo
• weka