Abstract
The Posture Video Analysis Tool (PVAT) is an interactive
Macintosh menu andbutton driven SuperCardTM prototype
for classifying working postures from video
footage. Following preliminary evaluations, a usability test
was conducted to test interface design and to identify
required modifications tothe software. Five users
participated in the study. Each user completed training (to a
preset criterion), test, and a post-test questionnaire. All the
sessions were video taped for detailed analysis. Preliminary
results indicate thatPVAT was acceptable in terms of setup
and video monitoring procedures as wellas screen layouts.
Detailed data reduction and analysis are in progress.
KEY WORDS:
Video analysis, userevaluations
Introduction
The Posture Video Analysis Tool (PVAT) is being
developed by the HumanFactors and Ergonomics Laboratory
(HFEL) at NASA Johnson Space Center in response to a need
for a low cost, reliable method of collecting postural data
from non-scientific mission footage. In the past, traditional
paper-and-pencil analysis of mission video footage revealed
that important information such as duration and frequency of
different postures could be extracted from video with the use
of a uniform methodology. Several systems exist for
evaluating working posture [1] [2][3] [4] [5]. These types of
tools are ideal when the evaluator has control of the data
collection procedures. They require that specific reference
points to be identified prior to classifying the working posture
in addition to the necessity of predefined camera views.
Therefore, these tools do not meet the needs of the
HFELprojects. PVAT, on the other hand, provides the
structured methodology needed to extract and
classify working postures, even from videos not recorded
specifically for experiment alanalysis.
TOOL DESIGN
PVAT is an interactive Macintosh menu and button driven
SuperCardTM software prototype consisting of several
input/output screens. It allows the user to define and select a
set of options relating to posture and tasks. Prior to the video
analysis process, the user must first define the posture
and/or activity parameters. These parameters are entered
using keyboard and pull down menu selections from the
PVAT"STARTUP" screen. For reference purposes, the user
can enter the particular video footage title, task description,
rater code, subject code, and camera location/view. The
primary posture parameters are: body orientation, body part,
body movement rating,and activity types. Body orientation
describes the overall posture most often observed in
microgravity (i.e., neutral body, horizontal, vertical,
and transverse). Body part (e.g., Neck) may be selected as an
additional descriptor variable. Body movement rating is
associated with dynamics of the selected body part
(e.g.,neck flexion versus neck extension) relative to the
nominal position. To further describe the target posture, a
rating level must also be selected such as severe versus mild
flexion.The rating level input is made via a pull down menu.
A final option available is behavior or activities which consist
of task components such as grasp and reach.
There are three analysis modes: 1) Body movement rating
only, 2) Activitytype only, 3) Both body movement rating
and activity. The body movement rating mode pertains to
specified body parts and their movement. Activity type mode
may include evaluation of either the task or overall body
posture categories. Finally, the combined mode involves the
evaluation of body movement (primarily) along
with activities (secondary).
Once all of these entries have been completed, the analyst
presses the"INITIALIZE" button to begin the analysis,
where the "INPUT" screen appears. Other features of the
PVAT prototype include: a terminology and definitions
library; a brief animation illustrating selected
postureclassifications; on-line help; print report capability;
and basic data reduction summaries (i.e., frequency and
duration). Once the setup is completed, the user may begin
analyzing the selected video footage. At present, the
software does not provide video control. Each
posture classification is stored and time stamped
automatically. Having completed the assessment process,
the user cansummarize the data for posture classification,
frequency, and duration usingthe "CALC" button on the
"INPUT"screen.
PHASE I - PRELIMINARY EVALUATIONS
Three iterative user evaluations were performed to identify
the interface design problems and required modifications to
the software prototype. The evaluations pertained to
theanalysis of various pre-selected video footages of crew
and non-crew operators working at two different workstations
during previous Space Shuttle missions or NASA's KC-135
reduced gravity aircraft. Two users were trained in each
test.They were tasked to view the video and time stamp the
selected posture categories. At the end of their analysis, they
provided comments on the PVAT interface through a
questionnaire. Results of thesepreliminary evaluations
indicated no major problems. However, some
minormodifications were required to simplify the interface
and improve the realstate utilization. For example, three
different input screens were created for different modes
ofanalysis. In addition, some of the buttons were renamed.
A "CALCULATION" option was added to provide a quick
summary of each run in terms of frequency and percentage
time for each posture category.
PHASE II - USABILITY EVALUATIONS
A usability evaluation was conducted to test the interface
modificationsincorporated during PHASE I prior to its
release for general use. The primary objective of this
evaluation was to check out the interface designfor
simplicity and to make improvements so that a novice in
ergonomic evaluations could use PVAT easily.
Five users were trained to use PVAT and what to observe for
the evaluations.All of the users had Macintosh experience
(daily use). However, they were novice-in-ergonomic
evaluations of posture.
The 10-minute video segments were randomly selected from
an ergonomicevaluation of a lifting/ translation task
conducted by HFEL onboard theKC-135 aircraft. The test
conditions included three analysis modes. The experiment
task consisted of setting up the tool, analyzing the video and
documenting the summary data for each individual run. Each
user viewed the video twice. The posture-related categories
were pre-defined and explained to the users during the
training. Their level of learning was measured through
apractice run and a quiz. Following the familiarization with
PVAT, each subject completed actual video analysis
sessions. At the end of each session, the users were given a
questionnaire to document their comments on the
"STARTUP" andthree "INPUT" screens. In addition to
subjective comments, the completion time for10-minute
video analysis, as well as the user errors were collected
asperformance measures.
Preliminary results indicate that PVAT was acceptable in
terms of setup and monitoring procedures as well as button
labels and locations, and screen layouts. However, theusers
required status feedback for some of their actions. Detailed
analysis and results will be includedin the poster.
CONCLUSIONS AND FUTURE PLANS
PVAT is a tool intended to be the initial step in identifying
"limiting postures" and related workstation design
concerns. Furthermore, it may provide the ergonomist with
supporting data to specify "adequate"postures or "safe"
duration for potentially dangerous postures. Results of
usability testsconfirm that the PVAT interface was
acceptable. PVAT can also be used by non-aerospace
industries with little or no modification. Future plans are
totest PVAT applicability in commercial industry (e.g.,
health care systems) and release it for general use.
References
[1] Corlett, E. N., Madeley, S. J. and Manenica, I.
(1979).Posture Targeting: A Technique for Recording
WorkingPostures. Ergonomics, 22 (3),357-
366.
[2] Fisher, W. and Tarbutt, V.(1988). Some Issues in
Collecting Data on Working Postures.
Proceedings of the Human Factors Society 32nd
AnnualMeeting. Santa Monica, CA: Human
Factors Society.
[3] Karhu, O., Kansi, P. and Kuorinka, I. (1977).
CorrectingWorking Postures in Industry: A Practical
Method for Analysis. Applied Ergonomics, 8,
199-201.
[4] Leonard, J. and Keyserling, W. M. (1989). A Method
toEvaluate Neck and Lower Extremity Postures Using
Simulated Real TimeAnalysis. In Mital, A. (Ed.),
Advances in IndustrialErgonomics and Safety
I. Great Britain: Taylor and Francis.
[5] Priel, V. (1974).A Numerical Definition of Posture.
HumanFactors, 16, 576-584.
