Abstract
We report an empirical study of aircraft maintenance
workers using wearable visual interfaces and collaborative
systems to support troubleshooting and repair work.
Preliminary results suggest gains in coordination and ease of
work when users have shared hypertext, video and audio
capability. Study results are being used to inform design and
rapid prototyping of wearable systems for use in vehicle
maintenance systems.
Keywords
Wearable computers, empirical studies, collaborative work,
vehicle maintenance.
Introduction
Having wearable computer systems and telecommunications
that allow field workers to access information and contact
experts would be valuable in many settings, from airline or
other vehicle maintenance, to rural/home health care, to on-
the-job training for telephone repair people or disaster
assessment workers [1]. Mechanics working on an airplane
or truck soon may be able to clip on a belt with a very small
(1.5 pound) wearable system with 486 processing capability,
2 PCMCIA slots, speech input, wireless audio, and a head-
mounted visual display with VGA resolution to find the
current procedural and parts information while working with
their hands free to return the vehicle to service rapidly.
Collaborative wearable systems are expected to improve
organizational effectiveness by: 1) spreading organizational
expertise among workers, 2) providing fast access to
procedural, process, and schematic information for problem
solving, 3) supporting process re-engineering, 4) and
improving organizational memory. Whether field workers
coordinate their work and gain from using information
technologies in ways similar to white-collar teams is not
known [2].
Designing these systems requires more than simply
miniaturizing components and building radio networks.
Designing systems so that they improve task performance, so
that displays are easy to read and hands free input is easy to
perform, so that work teams can share information, and so
that the technology fits into the organization requires a
diverse program of research on human-system interaction.
A review of the recent (1992-1994) CHI and CSCW
Conference Proceedings reveals few empirical studies of
the impact of information technologies on blue collar and
professional workers who work outside of the office and
with physical objects. At Carnegie Mellon University, an
inter-disciplinary team of engineers, computer scientists,
industrial designers and behavioral and social scientists
began experiments and field studies in 1994 to explore
some of the HCI issues mentioned above and in the
interactive poster "On Site Maintenance Using a Wearable
Computer System".
This initial empirical study focuses on a system to support
aircraft maintenance personnel when they inspect and
make decisions regarding repair or replacement of
bearings for an aircraft propeller.
METHOD
Subjects
Eighty-four volunteer participants were recruited from the
student body at the Pittsburgh Institute of Aeronautics
(PIA), the regional center for mechanics preparing for
careers in aircraft maintenance. Participants followed PIA
procedures for inspecting the bearings in a propeller and
making a decision about whether to service or replace the
bearings. All but one participant was male and their
average age was 23.7 years. Students had an average
grade point of 3.4 (on a scale where 4.0 = A). On average,
they completed 5 quarters of a 7 quarter-long educational
program. Seventy-nine percent of them had training and
experience working in the field of aircraft maintenance
while in military service prior to enrolling at PIA. They
rated their prior experience with computer use as an
average of "3" (on a scale where 1 = none, 4 = moderate
and 7 = Extensive). Also they rated their skill level for
repairing mechanical equipment as an average of "5.3" on
the same scale.
Procedure
Three variables of interest we focused on for this study
are: 1) use of paper manuals versus an on-line, hypertext
computer manual 2) task performance solo or with a
collaborative, remote helper, and 3) the presence or
absence of video links between the fieldworker and helper
so that the helper could see what the field worker was
doing. Study participants were randomly assigned to
treatments. All participants used the same apparatus and
worked in an aircraft hanger at PIA doing the experimental
task (see Figure 1 below).
FIGURE 1: Photograph of student wearing apparatus
and doing the experimental task
Each participant donned a pair of Virtual Vision glasses
with a miniature video camera attached over the right
eyeglass area. They wore the glasses and an audio headset
for the duration of the activity (participants took between
15 and 45 minutes to complete the task). The visual
display used had NTSC resolution. The hypertext aircraft
manual and parts information was done in 14 to 18 point
fonts.
Measures
At the beginning of the experiment participants completed
questionnaires providing demographic data and they took
short tests of verbal and visual ability [3]. They also
completed post-trial questionnaires describing their
experiences using the wearable system. Video and audio
taping was done to collect think aloud protocol data and a
brief exit interview was done to ask about user preferences
for future systems design. Participants rated their
perceptions about ease of finding information, seeing text,
seeing diagrams, seeing the workspace (interest was in
peripheral vision and seeing the area around the person
while wearing the Virtual Vision glasses), comfort in
wearing the glasses and in wearing an audio headset -- a
scale of 1 = very easy or very comfortable to 5 = very
difficult/uncomfortable was used for these items. Also,
participants were queried about the ease of hearing the
remote helper, being heard by the helper, and being
understood by the helper. Outcome measures, e.g.,
participants' time to complete the task and accuracy of task
completion were measured and will be reported when data
analysis is completed.
RESULTS AND CONCLUSIONS
This study provided insights about several basic usability
challenges and about coordination of work with wearable
systems. Self-report and video data provided usability
feedback especially for visual display requirements and
hypertext navigation and chunking constraints. Eye
accommodation was a serious problem for participants
(and would be more acute for older users). Seeing the
text in the on-line application was a serious problem. Both
screen resolution and screen "real estate" led participants
to rate on-line conditions on average as relatively more
difficult than the paper options (X2=31.031,p=0.058).
Similarly, participants indicated that seeing diagrams was
easier in the paper condition (2.5 vs. 3.9 for on-line
conditions ). Moreover, premature use of hypertext
linking options seemed to lead solo users to get lost in
details before grasping the overall sense of the task and
procedures.
Preliminary analyses suggest that doing this task with a
more experienced helper with the shared video to support
coordination of work is the most effective treatment.
Sharing a view of the work area helped the team to
coordinate their conversation effectively [4]. Additional
analyses of both process and outcome data are underway.
The video data show novices relinquishing the cursor to
helpers for hypertext navigation and helpers knowing
immediately which procedural step or part to display to
assist the novice in task completion. Self-report data
indicated that hearing a helper was rated across all
treatments as 1.2 and being heard by the helper was
similarly easy (1.04). Finally, participants rated their
ability to get the remote helper to understand the parts they
referred to as 1.5 (where 1 = very easy).
Results from this study are being used to inform system
design for wearable computers under development in
CMU's Engineering Design Research Center and in
industry. The research issues mentioned in the
introduction are addressed in this and other studies. This
experiment enables us to examine individual and
collaborative usability issues including current
communication patterns and aircraft maintenance
performance now and with prototype wearable systems.
ACKNOWLEDGEMENTS
This study was conducted with exceptional support from
the following institutions and individuals: Pittsburgh
Institute of Aeronautics, especially Arthur Adams, Rob
Fish of Bellcore, Bob Earp, Richard Martin, Dan
Siewiorek, Mark Miller, David Kaplan, Bethany Smith,
Elise Nawrocki, and Sandy Esch.
References
1. Smailagic, A. and Siewiorek, D. A case study in
embedded systems design: VuMan2 wearable computers,
IEEE Design and Test of Computers, 10,4 Sept. 1993.
2 R. Kraut, R. Fish, R. Root, and B. Chalfonte. Informal
communication in organizations: form, function, and
technology. In R. Baecker, editor, Groupware and
Computer-Supported Co-operative Work, pages 387-314,
Morgan Kaufmann, 1993.
3. Eckstrom,R.B., French, J.W., Harman, H.H., & Dermen,
D.. Kit of Factor-Referenced Cognitive Tests, 1976,
Educational Testing Service, Princeton, New Jersey.
4. Clark H.H. & Brennan, S. (1991) Grounding in
communication, In L. B. Resnick, J. Levine & S.D.
Teasley, Eds. Perspectives on socially shared cognition.
Washington, DC: APA Press.
