| |
Here, you find a survey of the main problems experienced
and their potential solutions based on our experience with
the ASL Model 501 mobile eye tracker gained in the studies
performed.
To help you
understand the tips and tricks, this is a picture of the mobile
eye tracker's head mounted optics we are using and it's labelled
components:
- In general:
- The calibration environment is optimized during the
procedure by minimizing excessive overhead light, sunlight,
and/or spot lighting resulting in better pupil and corneal
reflections.
- Preparation:
- Preparation should include a few pre-studies done
with various people to understand:
- Where the best lighting conditions exist
- What is the best subject positions are with the
mobile system(s) to be studied
- What is the best way to secure the user for the
calibration process
- It is a good idea to evaluate the study environment
at least daily while performing studies over longer
time spans. This is because conditions may change daily
in the environment.
- Calibration:
- Calibration surface:
- It is best to use a neutral, non-glare surface
- Should be located in a position covering the subject's
primary field of view in the desired scene plane
- Should only be a bit larger than the actual scene
of interest
- Should normally be marked with nine fixed points
covering most of the scene camera (see picture above)
image
- e.g.
- When placing the calibration screen, it should
be located in a similar position as the actual scene
plane in order to retain accurate calibration
- Perform the calibration quickly, for the reason
that subjects can become fatigued during the process
- Quick calibrations minimize the likeliness
that the subject moves during the calibration
process
- When the calibration is complete, the calibration
surface may be removed or replaced with the desired
(mobile) system.
- Use of different-sized screens:
- According to the size of the actual scene of interest
- For example, when the scene of interest is
the size of a PocketPC, then the calibration
screen should be a little smaller than an A4
sheet of paper
- Stool:
- During calibration, restrict physical movement by
holding the head/body of the subject as much as possible
immobile
- We found it extremely beneficial to help the
subject accomplish this by having him/her sitting
on a stool
- Chin rest:
- It is extremely important to find an adequate method
of temporary head stabilization during both the set
target procedure and calibration data entry procedure
for accurate data recording results
- To ensure the subject does not move, it is important
to stabilize him/her through the use of a chin
rest
- We use the Cheekrest
offered by ASL (right hand side)
- Tripod:
- To fasten the Cheekrest, we use a tripod with adjustable
height
- It can also help the subject to stand still by leaning
slightly on the Tripod (an alternative to the stool)
- Video recordings:
- Of the scene camera image including the crosshair
during the experiment for later analysis
- We use Camtel's USB TV Genie
- Use an encoder (e.g. avi2vcd) to turn your larger
avi files into smaller mpeg files
- Data analysis:
- Event tables :
- Each event (see below) was logged systematically
according to begin and end time in a table
- An event is defined as an action the user is performing.
- For example, an event can be defined as walking
or walking again if the user has previously
stopped walking.
- Example event table (click on the image to
enlarge it)
- This data is investigated to understand the user
behaviour in the context of the developing PERT
diagrams
- PERT diagrams:
- Program Evaluation Review Technique (PERT) diagrams
or Critical Path diagrams
- Example PERT Diagram
|
|
