Hi, I am running a project that uses the Myo armband to control several servos. I have noticed that sometimes the Serial read on line 124 of the pyomyo.py file does not complete or timeout and therefore the program hangs.

Perhaps setting a timeout on the Serial may help avoid this?

• Added simple classes for live labelling and classification for:
  KNN, SVM, Decision Trees, XGBoost and Logistic Regression Classifiers
• Added a deletion function to erase data labelled using a classifier.

Note these only work with the preprocessed emg mode, however its unlikely a classifier would work well on unpreprocessed data anyway.

PR is a work in progress, as I hope to clean up the pygame plotting and move a classifier into pyomyo and update the version on PyPI.

Using an enum for the EMG mode so people don't need to remember the difference between 0x01, 0x02 and 0x03. Changing the class name from MyoRaw to Myo.

Sensor positioning is hard and solving it is part of solving cross session generalisation, although there are fancy techniques that can be performed after the data is gathered, it would be worth having a discussion on the simple things. Here are some of my rough thoughts:

Calibration Methods

Once a placement is chosen, originally I would sharpie an outline of the whole Myo on my hand. It's important to mark the position of each sensor pod, not just one. The Myo can both rotate around the arm, move up and down, or tilt up and down.

One simple way I calibrate is by placing the Myo on my proximal forearm of my right hand, waving my wrist right and then making sure this movement only peaks one channel, e.g. channel 3. If the movement peaks two channels, I know the relevant muscle is in-between 2 EMG sensors and then rotate the Myo slightly to minimise cross talk and maximise the amount of the signal only picked up by channel 3.

Once this placement was used a kNN classifier can be trained on relevant gestures, e.g. 5 finger flexion. Then next session, after manual calibration is attempted it can be fine tuned until the kNN classifier works.

Search for something better, or at least more scientific

Some tools could be made to tell the user to rotate the Myo and show the user how similar the readings are to the previous data gathered.
Adding live PCA plotting to the live classifiers with a print out of between/within group variance may help quickly tryout different positioning for different gestures.

Both feature engineering and preprocessing of EMG data will likely be added for any significantly difficult project using pyomyo. If so it would be nice to contribute these implementations back so that efficiency improvements/bug fixes can be shared. It would be nice to include a metrics.py file containing implementations and a plot_metrics.py file that can give some intuition on what each feature captures.

Some examples of metrics to include are Mean Absolute Value (MAV), Root Mean Square (RMS), Willison Amplitude (WAMP), Waveform Length (WL) and Zero Crossings (ZL). More info can be found in this paper for different metrics. My hope was to implement them and create a notebook showing their feature importance's for different tasks and models but have not had the time so far.

This was started in the Spectrogram branch and is completed for one channel, although I am not sure about the correctness of my method. Needs moving to all 8 channels and rebasing.

The BLED112 dongle allows the Myo to work on any operating system as it acts as a COM port that forwards serial commands send to it over Bluetooth.

One reason why it's worth doing this is due to the poor cross platform BLE support, however bleak (Bluetooth Low Energy platform Agnostic Klient) may solve this and allow people without a dongle to use pyomyo.

This feature would also act as a stepping stone for makers of open source BLE EMGs to add support for their hardware to pyomyo.

Thankfully, Thalmic Labs released (most of the) the specification here.