cog-ethz / evertools Goto Github PK

Currently the position offset of all outputs is returned giving you values from 0 to n-1. This should be correctly mapped to output.

Create helpers that load a specific sensor's data by name.

@raweibel I don't have any session parameters in the test database, I will not update the function right now. If you can give me a session with parameters I can add it, or you can fix this up after I finished the other functions.

There is an issue with the alignments of offsets in EVE cog-ethz/EVE_dev#113 that currently impedes the questionnaire for boolean values to be read out properly.

Provide for a participant the propensity to look around (PLA) compared to completing the path. The underlying assumption (due to control constraints) is that if people walk they are on "path" and that if they look around without moving they "explore". We compute PLA by comparing the view direction with the path direction. To account for the symmetry of the view direction (i.e. negative and positive angles could cancel each other out), we take the absolute view angle.

Since the view direction is 3D, we will use quaternions to sum all angels before normalising. The final quaternion will be used to derive the angle between the unit path direction vector and the averaged view direction.

Additionally, we can compute a standard deviation from the final view angle to each individual view angel to assess how much participants PLA varies.

We will use RSpincalc for quaternion computations.

Same as in #4, but for orientation

Create helpers to gather time intervals for scenes in a session (to filter sensors).

Several questions have multiple text fields to fill out, using tab to switch through labels would be useful.

This could also be extended to go through choice options and Back/Next button.

Slight changes have to be made to comply with the new question format.

To test whether the normality assumption is violated, instead of only plotting the diagnostic plots, we can run crTest from DAMisc to compare the LOESS curve with the optimal PP/QQ line. This provides a significant p-value, if the normality assumption is violated in a way that the distribution significantly differs from the theoretically expect PP / QQ plot.

Create helpers to get scenes by name or id.

Code needs to be upgraded for evertools to work again.

Instead of reading the position table, we now load the sensor name and then the 3d data.

in cog-ethz/EVE_dev#113 we fixed the reason for the offset between the two tables. Currently the sensor results will be jumbled with read out. FIX ASAP.