/SpatialTaskV2

Code for SpatialTaskV2

Primary LanguageMatlab

SpatialTaskV2 - draft

Requirements

Instructions

  1. Set up Linux/Python computer and eye tracker
  2. Set up Windows/MATLAB computer and task
  3. Introduce subject
  4. Calibrate and start eye tracker recording
  5. Start task
  6. Monitor task
  7. Save data and Shutdown

1. Set up Linux/Python computer

  • Open terminal.
    • cd anaconda2
  • Run Spyder (Python 2.7).
    • Open EyeTracker/Run.py.
  • Open eye tracker software.
    • Turn on detection of both eyes.
    • Check pupil remote plugin is loaded
    • Find server port number under "Pupil Remote" menu.
  • In run.py in Spyder:
    • Set port number for the pupil remote server (port).
    • Set address and port of TCP server (TCPAddr, TCPPort). Address should be Linux computers adress on network.
    • Set filename name for pickle file (use subject ID).

2. Set up Windows/MATLAB computer.

  • Check cable 7 is plugged in at back of MOTU and cable 16 isn't.
  • Turn on MOTU. Check that sampling rate indicator light on front settles on a value. If it doesn't, restart the Windows computer.
  • Turn on both amplifiers.
  • Run testSpeakersLights.m to check all speakers and lights are working as expected. Each speaker/light pair should come on one by one.
    • If multiple speakers/lights don't work, check amplifiers are both on.
    • If light/sound are sperated, check cable mapping.
    • If audible clicks, LED output is going to a speakers. Check cable mapping (specifically switching channels 7 and 16).
  • Set up touch screen.
    • Check it's plugged in.
    • Check it's virtually located to the bottom right of the main monitor.
  • Open MATLAB 2013b.
    • Set working directory to C:\Gareth\SpatialTaskV2\MATLABTask.
    • Open run.m in editor.
    • Set temporary subject ID
    • Set TCP address and port (params.TCPAddr, params.TCPPort) to match Python values.
    • On Python computer click run in run.py
    • On MALAB computer click run for run.m
    • Check touchscreen figure is aligned with monitor.
    • Check touchscreen figure calibration is correct (do the plotted dots align with the red dots?) If yes, enter 'y' and press enter.
      • If not:
        • Press crtl+c to stop the task.
        • dbquit if necessary.
        • Enable calibration in run.m's parameters.
        • Run run.m and re-calibrate.
    • Check the touchscreen's touch calibration is correct (does pressing on the screen click in the correct place?).
      • If not, double click eGalaxTouch icon in system tray and run monitor mapping.
      • Check touchscreen again, if calibration is still wrong, right click eGalaxTouch icon and run 4 point calibration.
    • Enter chamber and check stimuli are working.
    • End test by pressing ctrl+c.
      • Type dbquit if in debug mode (indicated by K>> in command window).
    • Set parameters for task in run.m.
      • Set subject ID.
      • Set nBlocks to ~8 (1 block = 100 trials = ~250 S).
      • Set nBreaks to ~4.
      • exchangeTime to 1.
      • PTSKip to 0.

3. Introduce subject

  • Give information sheet to subject to keep.
  • Get subject to read and sign consent form.
  • Explain eye tracker and basic task requirements.
    • Keep gaze fixated on target during stimulation.
    • Respond on touchscreen.
  • Explain paradigm.
    • Pre-training phase (<5 mins).
      • AV stimuli will come from same location, subject has to indicate loaction with one press on touchscreen.
      • They can look around after stimuli before responding to orientate themselves, but should look back to red fixation LED after responding. This will be tracked.
      • Pre-training will continue until a threshold of localisation accuray is reached.
      • When finished, a message will appear on touch screen saying ready to begin main task.
      • When ready, press screen again to continue.
    • Actual task (~1 hour).
      • Subject needs to keep eyes on fixation LED while stimuli are playing (~1s).
      • Subject needs to respond on touchscreen twice:
        • First to indicate auditory location.
        • Second to indicate visual location.
      • If the task stops, check touchscreen message to see if it's a break.
      • Touchscreen messages will guide subject with what to do.
  • Outside booth:
    • Place eyetracker on subject and position eye cameras.
    • Unplug eye tracker.
  • Move subject inside booth.
    • Plug eyetracker in inside booth and re-enable eye detection.
    • Check eye tracker camera pupil detection and adjust as required. Make sure detect pupil 0 and 1 are both enabled in pupil app.
    • Check world camera alignment.
    • Show subject expected fixation LED location.
    • Raise/lower chair so subject's head is vertically level with fixation LED.
    • Hand subject touchscreen. Make sure the cables don't drop out.

4. Calibrate and start eye tracker recording.

  • On the Linux computer select the eye tracker software. If it's already running, don't click the taskbar shortcut again.
  • In the world window:
    • Ensure the target surface markers are detected (green boxes around markers).
      • If not, adjust "min_marker_perimeter" in "Surface Tracker" menu.
    • Ensure surface is defined, detected, and named "Target".
      • If not defined, add and name as "Target" in Surface Tracker menu.
      • If not detected, check markers are being detected.
    • In the "Calibration" menu select "Natural Features Calibration".
    • Begin calibration by clicking (C) in the top left.
    • Ask subject to look at each world marker in turn and click to make these in the world.
    • Press C key to end calibration.
    • Check calibration accuracy.
    • Press (R) to begin recording video/gaze data.
  • In Spyder:
    • Click in run.py and click green run button.
    • Check Python connects to ZMQ server and waits for MATLAB to connect to TCP server.

5. Start task

  • On the Windows/MATLAB computer:
    • Run run.m.
      • Respond 'y' to calibration question.
  • On Linux computer
    • Check MATLAB has successfully connected to TCP server and Python has begun collecting data from eyetracker.

6. Monitor task

  • Tell subject to press touchscreen when ready to start (twice).
  • Monitor pre-training.
    • Check subject is looking at fixation light when trials start.
    • Subject should pass quickly, if not, re-explain task!
  • Monitor task.
    • Check subject is returning gaze to fixation LED/Target surface before each trial starts.
    • Hope Windows computer doesn't bluescreen randomly.
  • MATLAB task can be stopped with ctrl+c.
    • If task is stopped, or crashes for another reason, MATLAB enters debug mode.
    • If there is data that needs to be saved, save the workspace manually (data is only automatically saved during each break, but not after every trial).
  • When finished, untangle subject from cables and debrief.

7. Save data and shutdown

  • On Linux computer:
    • In world view click (R) in world window to stop recording.
    • In Python click stop to stop collection.
      • This raises an exception, which will be caught.
      • Python will then automatically convert the saved pickle file to a .mat file.
    • Copy the directory containing the eyetracker video from /Home/recordings/ to /Home/recordings/SpatialTaskV2 Recordings/.
      • Rename the backed up directory from the date to the subject ID.
    • Close eye tracker software.
    • Unplug eye tracker.
  • On Windows computer:
    • When the task completes, MATLAB saves its data automatically.
    • In ...\Data[SubjectID][Datetime]\TeComplete.mat.
    • Copy ...\Data[SubjectID]\ directory to external backup.
    • Close MATLAB editor (to prevent open files automatically re-opening next time someone starts MATLAB).
    • Close MATLAB.
    • Turn off amplifiers.
    • Turn off MOTU soundcard.

To do

  • Add live functionality
    • Via TCP server?
    • Minor modification to MATLAB code also required to monitor and act on online eye data.