VirteTodo

• See unfiltered Joystick and WIP data for Virte3 (and Virte4)
  • We should look at the printouts, identify a few common strategies, and categorize the trials by strategy.
  • We can determine which strategies the expert gamers tended to use and which strategies the people with the best overall task-performance used.
  • We can also look how the strategies of all the subjects changed over the course of the 18 targets (for v3) or the 3 trials (for v4). (This would be a first pass at identifying learning behavior.)
  • I think this is a relatively straightforward task, and it would make a great section for a journal paper.

• Matlab Mean-Tau visualization for virte3: analysis_2005/tau-plots/
  1. Make printouts of Tau & Velocity profiles for all 32×5×12 trials.
  2. Look at printouts to determine rejection criteria. I have a tool in matlab to view these plots one-by-one, but doing it on paper seems SO much easier.
  3. Make new CSV from the culled data-set, and delete the rejected trials from the existing XLS files that Sarah has already started to work on.
  4. Generate plots and printouts of the mean tau curves.

• Distribution of stopping-points in Virte3.
• Distribution of stopping-points in Virte4.

• Fix tau plots for the Virte3 Joystick data
  • In Virte3 joystick data, make a distinction between the initial stopping point (the point we use to line up the pagths from different trials) and the final stopping point (where the person ended up after all the corrective nudges were finished).
    • Either:
      • Let paths run all the way to the final distance and have an index to the stopping point.
      • Add fields: time_after, pos1_after, vel1_after, tau_after.
  • Right now the stopping point is chosen where the velocity reaches zero in the filtered data, but this is slightly wrong in some cases. Come up with a better way of picking the stopping point.

• Exposure on Virte3 data.

• Be able to view Virte1, virte3, and virte4 logs in MView.

• Clean out double-humps and other bad trials in Virte3
  • Make all the print-outs first.
  • Document each rejection.
    • Which trials are missing to begin with.
    • Which ones i threw out and according to what criterion.

• Virte1 data
  • Remove the aborted trials.
  • Make CSV of per-path measures

• Understanding Tau (read papers!)
  • Braking with space in front of the observer.
  • 3 Lee papers (is there one about braking in a car?)
  • What happens if look and move directions are different.
  • What is Tau/TTC for points off the look or move directions (things passing near the observer without colliding).

• Tau visualization in MView
  • Visualize tau only within a ray in front of the cursor.
  • Do this for Virte3, Virte4 data.
  • Do this live in Virte4 enviroment, doing real-world, vr, joystick

• SVN check-in of Matlab scripts and data-sets. This will have to wait until my TortoiseSVN problems are solved.

• IRB training
• PCA on PerPathMeasures matrix
• Factor analysis on PerPathMeasures

Collaborative Tasks

• Have a log-viewing session of Virte3 or 4 data.

All combinations of tau and mean-tau plots

sub,      target,   condition
------------------------------
mean      mean      overlay      do this first

mean      overlay   ---          
mean,     ---       overlay      
overlay   mean      ---
---       mean      overlay
overlay,  ---       mean         skip this
---       overlay   mean         skip this

---       ---       overlay
---       overlay   ---
overlay   ---       ---


Could show STD and MEAN seperately. on the same axis.  one below the other.