THESIS
2021
1 online resource (xxi, 224 pages) : illustrations (some color)
Abstract
Fitts’ law is very robust and used extensively in human-computer interaction (HCI). It predicts
the movement time (MT) of pointing tasks under various conditions. However, the impact of
visual inversion on hand movement performance has not been quantitatively investigated. This
is really of essence as it provides a means to “unlearn” what we know and thus investigate how
the eye and hand coordinate works. This thesis comprises three experiments with three
different factors to investigate their impact as well as the underlying mechanism of aimed
movement. The first and the third experiments were related to visual inversion. In order to
determine the dominant factor of distal-pointing, the second experiment was designed.
The first experiment focused on the impact of eye-hand incompatibili...[
Read more ]
Fitts’ law is very robust and used extensively in human-computer interaction (HCI). It predicts
the movement time (MT) of pointing tasks under various conditions. However, the impact of
visual inversion on hand movement performance has not been quantitatively investigated. This
is really of essence as it provides a means to “unlearn” what we know and thus investigate how
the eye and hand coordinate works. This thesis comprises three experiments with three
different factors to investigate their impact as well as the underlying mechanism of aimed
movement. The first and the third experiments were related to visual inversion. In order to
determine the dominant factor of distal-pointing, the second experiment was designed.
The first experiment focused on the impact of eye-hand incompatibility and investigated the
substructure of the complete movemenet to investigate the presence of a ballistic region when
task difficulty is low. For this purpose, MT was separated into initiation time (IT), distance
covering time (DCT), and acquisition time (AT). The AT gives a better separation of the ballistic
and visually controlled regions on the index of difficulty (ID) than movement time indicating
the existence of a ballistic region.
The second experiment focused on pointing at different depths. The visual angles of movement
amplitude and target width are better predictors of movement time than the linear dimensions
and depth does not have a significant impact. With increasing depth, hand and arm tremor
effects increase.
The third experiment noted that the individual kinematic perception is correlated to one’s MT
and people do not need to see the hand during the whole movement when aiming. What we
need is a window nearer to the target. The duration and endpoint variability of the first ballistic
submovement of the first and the third experiment were compared and that helped explain the
performance difference among the different conditions. This validates that different visual
conditions affect not only the total MT but also the submovement properties. In addition, the
comparison turns out that people can perform delicate control strategies viewing direct, and
this ability is heavily reduced with screen feedback.
Further analysis using multiple datasets verified a few hypotheses and gives better models
without overfitting of data. Using data from the three experiments and several published studies
proved that the log
2MT has a better normal distribution than MT. Moreover, the use of log
2MT,
angular A and angular W, and two-part variations of original Fitts’ models were proved to have
more accurate prediction on MT. Besides, the model MT = a + b ∗ ID + c ∗ Gain/W, is
consistently valid over aimed movement with multiple depth and control-display gain (Gain).
Post a Comment