This implements the core functionality for the automatic evaluation of TUG-Tests (Timed Up and Go test, see here).
This implementation uses machine learning for the classification of the different phases of the test (stand-up, walk, rotate, sit down). The model needs to be trained before usage. A pre-trained model can be derived from the LabeledData. The server expects a live data-stream on a rabbit-mq server, which is produced by the corresponding android app.
A short overview (in german) of the project is provided in the overview.pdf.
Contributors: @hwiards, @cjmoeller, @RollingRhino