Skip to content

Latest commit

 

History

History
137 lines (96 loc) · 3.52 KB

TODO.md

File metadata and controls

137 lines (96 loc) · 3.52 KB

OpenBCI Driver

  • Automatic restarting if stream stops
  • Handle missing packets
  • Switch off common reference (noise)
  • Impedance checking GUI
  • Unit conversion to µV

Firmware rewrite (?)

change samplerate

Live Control Features

Training frequency selection widget

Recording Tool

automatic log file naming (only specify username)

Filters and such

  • Graphic autoscale/normalizing feature (based on marking a good range)
  • Peak filter
  • Find out of IIR filters work, and use correctly (maybe higher orders will work then?)
  • Artifact Rejection

Traits semantics

  • Disallow assignment of undeclared traits
  • Input trait for non-signal stuff, default value
  • constructor arguments, automatable? (for using init())
  • process call does not work for collections of Input()
  • Simplify threshold into one mode, with 0 treshold deactivating sides
  • set colors on filters
  • Threshold.passfail should be standard output

Towards Domain Specific Language:

*) Think thoroughly about semantics of Signal - Sample rate, nyquist - Problem field of 'shadow' signals, that have same content but different metadata - combine append/process

*) Reverse Control flow, start processing from output blocks Look at pyo, visitor pattern

- Enumerate all output Blocks
	* Output blocks register at some instance (Context)
	* Output blocks derive from a class OutputBlock
- Find out how much samples are to be processed
- Walk through the chain and process
- Cache results that are in path for other output blocks

*) Latency Measuring Test Case *) Threshold block with widget *) Module structure, naming *) Fix filter buffer size

Dependencies

  • Move away from rtmidi2 to something more standard, maybe pypm
  • investigate chaco to replace pyqtgraph
  • take a look at pyface

Design Decisions:

  • OutputPort's are buffering stuff themselves, as requested by InputPort's on connection

  • Source Blocks (Channels) can be requested through the Context

    • Sources are just a output port, no special class
    • Context has the source ports
    • Sources are identified through names, like 'Channel 1'

  • There is a device class, that controls the hardware driver, and can log data for replay/pause/seek functionality

  • Device and Context are bound together via signals

  • Recording class is named 'History'

TODO:

  • Limit Error message (and trace() feature)
  • Make a simple threshold widget
  • Buffer length request feature for Input() port
  • Give blocks a name, so they can store their config in a separate file in Python object notation

Recording Features

  • GUI Frame with notion of a "session", and record/playback functionality
  • Start/Stop/Pause and maybe looping
  • Freely usable WAV file record and playback blocks
  • Buffer reset/refill feature
  • Support for seeking
  • Spectrogram widget
  • Use WAV files everywhere

DSP:

  • Wavelet transforms for ECG analysis

Widgets:

  • Scrolling Plot / Oscillosope
  • Fire animation

Hardware:

  • Light & Sound Box (Grow light)

DSL Features:

Blocks:

  • Source
    • Wraps OpenBCI and WAV replay (Extra class/instance option for explicit replays)
  • (Random Noise)
  • Think of a way to register 'virtual' sources to Context

Transform:

  • Filter (configurable between IIR and FIR)
  • BandPass, Lo/Hi-Pass, Notch
  • Threshold
  • (FFT)
  • (Wavelet transform)

Output:

  • single buffer across filter chain (an idea from pyo, allow in-place processing)

  • Signal class:

    • Controls buffer length
    • Transports side channel data (for example Threshold)
    • Can transport transforms (frequency domain)?

  • Save configs to .config file (with ConfigParser?)