|
|
Louis Rosenberg Ph.D., Scott Brave
Immersion Corporation
2158 Paragon Drive
San Jose CA 95131
408-467-1900 info@immerse.com
This project uses a force feedback joystick to enhance
user interaction with standard graphical user interface paradigms.
While typical joystick and mouse devices are input-only, force
feedback controllers allow physical sensations to be reflected
to a user. Tasks that require users to position a cursor on a
given target can be enhanced by applying physical forces to the
user that aid in targeting. For example, an attractive force field
implemented at the location of a graphical icon can greatly facilitate
target aquisition and selection of the icon. It has been shown
that force feedback can enhance a users ability to perform basic
functions within graphical user interfaces.
Force Feedback, Haptic Interface, Manual Performance
This project focuses on using force feedback technology
to enhance user performance in manual tasks. This study employs
a joystick controller equipped with force feedback capabilities
to enhance user interaction with standard graphical user interface
(GUI) paradigms. While the work presented here has potential application
to improve performance for all computer users, the motivation
for this effort has been to develop a technology to assist persons
with neuromotor disabilities in interacting with GUI based computer
systems. Persons who suffer from spastic hand motion or hand tremor
have a difficult time performing the manual targeting tasks required
of standard GUI paradigms. This project implements a force feedback
joystick to assist disabled persons in performing tasks such as
clicking on buttons or icons, selecting items from a pull down
menu, and positioning the thumb on a scroll-bar.
A force feedback system consists of interface hardware and a computation
engine. The interface hardware typically includes a mechanical
linkage in the form of a joystick or exoskeleton which couples
the human operator to a source of mechanical power-either electromagnetic,
electrohydraulic, or electropneumatic actuators. The computation
engine is a processor which monitors the dynamic motions of the
interface hardware through sensors in the mechanism and commands
forces to the user by controlling actuators. The computation engine
governs the forces felt by the user through control algorithms
computed as a function of the sensor measurements.
Figure 1, User at Force Reflecting Joystick
In recent years, a number of researchers have explored the use
of force feedback technologies to enhance human performance in
manual tasks. Rosen and Adelstein have demonstrated that abnormal
human tremor can be suppressed using force feedback hand controllers
[1]. Repperger has shown that active controllers can effect both
performance and learning period in stick based manual tasks [2].
Rosenberg, has shown that force feedback can enhance manual performance
in virtual environments and telemanipulation systems. He developed
a general concept known as Virtual Fixturing in which force feedback
information is overlaid to assist user performance in a variety
of manual activities [3].
We have performed a controlled set of human tests to quantify
the effect that force feedback has upon manual performance in
fundamental user tasks within GUI computer interfaces. Our testing
paradigm includes three simple tasks: Targeting Buttons, Selecting
Items from Pull-Down Menus, and Positioning the Thumb on
a Scroll Bars. For each of these tasks we tested three cases:
No Force Feedback, Active Force Feedback, and Passive Force Feedback.
Thus a total of 9 tasks were performed as shown below:
| Button Task | Menu Task | Scroll Task | |
| No Forces | test 1 | test 4 | test 7 |
| Active Forces | test 2 | test 5 | test 8 |
| Passive Forces | test 3 | test 6 | test 9 |
Active and passive force feedback can be differentiated by whether
or not ENERGY is added to the system by the controller. Active
force feedback controllers apply forces to the user by adding
energy into the human-machine system. Passive force feedback controllers
apply forces to the user by removing energy from the human-machine
system. For example, an active controller might use servo motors
to generate feedback forces. The strength of the forces could
be directly regulated by the computer which regulates power to
the motors. A passive controller might use energy dissipation
elements such as a friction brake or a magnetic particle brake.
These devices can not directly apply forces to the user, rather
they can only apply resistance to the user's motion. The advantage
of active force feedback control is that it is inherently general.
When using active elements such as servo motors, the system can
produce any general force sensation. The advantage of using passive
force feedback control is that it is inherently stable and inherently
safe for the user. This is because energy dissipation elements
only resist motion but do not induce motion. Thus the tradeoff
between active and passive is a tradeoff between performance and
safety.
For active force feedback tests, attractive force fields were implemented on targets within the graphical user interface. In button trials, buttons were implemented with attractive force fields which captured the joystick. In menu trials, each menu element had an attractive force gradient that pulled the joystick to the center of the given menu selection. The resulting sensation could best be desribed as "snapping" from one selection to the next. In scroll-bar trials, an attractive field was implemented along the length of the bar, keeping the user from drifting off the bar while the button was held down. In all three cases, the attractive fields greatly reduced the targeting accuracy required of the given task.
In passive trials, attractive fields could not be implemented
because it requires adding energy to the system. Instead, capture-barriers
were implemented which would resist a users motion as they deviated
from a given target. In button trials, the passive resistance
would impede a user from moving away from a button once the cursor
came within a given proximity. In Menu trials, the passive resistance
would give a "bump" sensation as the user moved from
one menu item to the next. In scroll-bar trials, the passive resistance
would impede users from drifting off the bar once the thumb was
aquired. In all three cases, the passive resistance improved targeting.
A pilot study was performed on 5 test subjects. The mean results
of each test are shown below. Results indicate performance times
in seconds for completing each of targeting tasks. Low times indicate
fast task execution (high performance).
| Button Task | Menu Task | Scroll Task | |
| No Forces | 4.4 sec | 5.8 sec | 6.6 sec |
| Active Forces | 2.7 sec | 4.2 sec | 6.2 sec |
| Passive Forces | 2.9 sec | 4.8 sec | 5.9 sec |
The pilot study has shown that both active and passive force feedback
can be effective in decreasing task completion times (increase
user performance). This pilot study confirms that force feedback
is an effective means of enhancing manual interaction with graphical
user interfaces and suggests that additional tests should be conducted.