notes.txt

Workshop notes:

- Figure out your phone's approximate sampling rate:
    * Only collect one channel
    * matlab/python - get approximate sampling rate
- Download a file explorer app
- using LINEAR_ACCELERATION instead of pure ACCELERATION, why?

Files with TODO flags:
- Android:
    * GestureRecognizer
    * FirFilter
    * SensorReader


Rough outline:
- assemble into groups
- talk about how android sensor collection works
    * Using LINEAR_ACCELERATION instead of pure ACCELERATION, why?
- run CSV app, collect some sample data (comment out gyro)
- figure out approximate device sampling rate
- visualize raw CSV data in matlab (no resampling)
- Resampling
    * What is resampling?
    * Why do we need to do it?
    * What are a few different kinds of resampling? (ZOH, linear, etc)
    * GROUP ALGO: Write me a resampler
- Filtering
    * What is filtering?
    * Why do we need to do it?
    * (NOTE: Nyquist? Does that apply here?)
    * What does a filter look like (FIR block diagram)
    * GROUP ALGO: How do you do manual filtering in code given streaming data?
- Filter Design
    * How do you spec a filter?
    * How do you choose what filter type to use?
    * DEMO: trying multiple tap lengths
- Gesture recognition
    * Couple different ways to do gesture recognition:
        > ML style
        > DTW
    * Our way: thresholding
    * GROUP ALGO: What would this thresholding algorithm look like?
- Back to android:
    * Fill in all the TODO's back in the android code
        > This namely involves parameters, a few algorithm things
    * How did you do?