Najjar, L. J., Thompson, J. C., & Ockerman, J. J. (1997). A wearable computer for quality assurance inspectors in a food processing plant. In Digest of papers: First International Symposium on Wearable Computers (pp. 163-164). Los Alamitos, CA: IEEE Computer Society.
Lawrence J. Najjar
Georgia Tech Research Institute
GTRI/EOEML/MARC Room 335
Atlanta, GA 30332-0823 USA
gt4708d@prism.gatech.edu
J. Christopher Thompson
Georgia Tech Research Institute
chris.thompson@gtri.gatech.edu
Jennifer J. Ockerman
Georgia Tech Research Institute
jojo@chmsr.gatech.edu
We were asked to try to improve the performance of quality assurance inspectors in a food processing plant. These workers walk around a very large, very loud, football field-sized plant and take food samples at various processing points. The workers pick up a sample, make a measurement (for example, temperature), wipe their hands on a paper towel, then use a pen and clipboard to write the measurement on a paper inspection form. At the end of the shift, the quality assurance inspectors give their paper forms to the quality assurance manager. For other members of the management team (such as the plant manager) to see the inspection data, the forms must be photocopied or the data must be typed into a networked computer.
We faced several challenges in our effort to improve the performance of the quality assurance inspectors. The inspectors need a data collection system that is portable and allows inspectors to enter data without using their hands. The system must also provide the data to the plant management team as the inspectors actually gather the data. Finally, the use of sound to enter or receive information is a major challenge in the plant's noisy environment.
Based on an analysis of the inspectors' tasks and the system requirements, we decided that a wearable, voice-operated computer system with a wireless network would be the best solution [1, 2, 3]. With input from the inspectors and plant mangers, we iteratively sketched, prototyped, and evaluated three different system alternatives.
For the fourth, and current, version of our wearable computer system, we slightly modified one of the new, commercially-available, wearable computers. We purchased a system, but we replaced the two-hour-long batteries with four-hour-long, neoprene-coated, nickel metal hydride battery packs that we assembled ourselves. We also added PCMCIA cards for 16 bit, full-duplex audio and a 2 megabits per second, wireless network connection. The full-duplex audio allows the wearer to issue voice commands while audio information plays. The wireless network allows the wearer to send inspection data from anywhere on the plant floor to the managers' computers.
We included a microphone for voice input and an earphone for audio output. For hearing protection, we found small, inexpensive ear plugs that deaden all sounds but those delivered close to the wearer's ears by an earphone. We also used a light, dominant eye-adjustable, monochrome, 640 x 480 pixels, head-mounted display for wearer comfort. Figure 1 shows the latest version of our wearable computer system.
Figure 1. Our latest wearable computer system.
Our system also includes software for speaker-dependent, voice recognition, the wireless network, auditory feedback, our customized quality assurance inspection application, and the Windows 95 operating system to allow the applications to run simultaneously.
We designed a customized application for the quality assurance inspectors to use while performing their tasks with the wearable computer system. First, the inspector selects the inspection form with which the inspector wants to work. One of the most frequently used inspection forms is used to record the temperatures of various chicken meat samples (e.g., white meat, dark meat, tenderloins). The inspector says the type of meat being sampled (e.g., "white meat"), and the low and high temperatures of the samples (e.g., "3 4," "3 6"). The application shows the screen for that type of sample, enters the time, and displays the temperature values spoken by the inspector. The application automatically adds a comment when a value is outside the desired range. For example, Figure 2 shows the screen for the temperatures of white meat samples, including a comment that the first value is higher than the desired range. The system uses the earphone to repeat each recognized value to the inspector. To reduce the number of voice entries, the application automatically moves the cursor to the next entry field. The inspector can erase the last value, go to the "Tenderloins" or "Dark Meat" sections of the form, see a summary of all entered values, select a different form, or quit the application.
Figure 2. Application screen.
After the worker enters the values, the system uses the wireless network to automatically transmit the values to a central computer. To view the information, the plant managers access the central computer. We recently completed a successful field test of our current wearable computer and inspection application.
Designing, prototyping, and evaluating wearable computer hardware and software alternatives iteratively with users and managers is very helpful. Although the size and weight of wearable computer components are important, the wearable computer system components must either stay out or move out of the wearer's way. We can improve voice recognition accuracy if we use longer, highly differentiated, voice commands. Finally, the typical graphical user interface does not work with voice input. The older, command-based or function key-based user interface is much more effective. To learn more about our project, please go to http://mime1.marc.gatech.edu/mime/epss/default.htm.
[1] J. J. Ockerman, L. J., Najjar, J. C. Thompson, and C. J. Treanor, FAST:
A Research Paradigm for Educational Performance Support Systems, Educational
Multimedia and Hypermedia 1996 (edited by P. Carson and F. Makedon), Association for the
Advancement of Computing in Education, Charlottesville, VA, 1996, pp. 545-550 (also
available at http://mime1.marc.gatech.edu/mime/papers/edmedia2.html.
[2] L. J. Najjar, J. C. Thompson, J. J. Ockerman, and C. J. Treanor, Using a
Wearable Computer for Industrial Data Collection and Performance Support, paper
presented at the Workshop on Wearable Computer Systems Meeting, Renton, WA, August, 1996
(also available at http://mime1.marc.gatech.edu/mime/papers/Boeing.html.
[3] L. J. Najjar, J. C., Thompson, and J. J. Ockerman, Using a Wearable
Computer to Improve the Performance of Quality Assurance Inspectors in a Food
Processing Plant, paper presented at the CHI'97 Workshop on Wearable Computing,
Atlanta, GA, March, 1997 (also available at http://mime1.marc.gatech.edu/mime/papers/CHI97_position_paper.html.