Variability in Wrist-Tilt Accelerometer Based Gesture Interfaces - - PowerPoint PPT Presentation
Variability in Wrist-Tilt Accelerometer Based Gesture Interfaces Andrew Crossan 1 & Rod Murray- Smith 1&2 Hamilton Institute, NUI Maynooth 1 Department of Computing Science, University of Glasgow 2 Overview Rationale Mobile
Rationale Mobile Devices & Accelerometer
Targeting & Gesturing
Experiment Conclusions
How can we adapt the dynamics of our
A limited amount of screen space on
Low graphics resolution Small onscreen targets Interactions can be slow and error
With many emerging technologies, interaction with
(e.g. gesture recognition, haptic devices, speech recognition,
animation),
the user is in constant and closely coupled interaction with
the computer.
(as well as old-fashioned discrete communication such as
selection by pressing buttons or typing via a keyboard.)
Interaction takes place over a period of time - dynamics of
interaction are important, so a snapshot time sample is insufficient.
Variability is introduced into the system
Probablistic methods required
PDA with xSens P3C
Continuous control One handed potentially
Input mechanism and
Study examining wrist tilt as an input
Measuring variability in short direct
Is it equally easy to target in all
Task
Select & hover over highlighted
Alternate between centre and
12 participants (seated posture)
10 right handed & 2 left Using dominant hand
1 training session
6 times to each target
2 experimental sessions
12 times to each target
Marble metaphor
Cursor gain set
Target distance
48o in x 36o in y
Target widths
7o in x 5o in y
Slip off errors Excess path length
Above the minimum 85 pixels distance
Time to target Unintentional movement
Confusion with the mapping in 30 out
Bubble vs. marble on elastic metaphor Remove these from the final results
Easier to target in the lower half of the
Less variability moving to lower targets High error rate… but deliberately
Lower variability in Y while hovering ???
Presented a study examining variability in
Easier to targets in the lower half of the
Non linear cursor gain to compensate Future work
More realistic cursor gains Dynamic cursor warping Accelerometer gesture input Mobile situations