Dynamics of Supervisory Control Task Performance:
SCAMPI Project Summary Barrett S. Caldwell University of Wisconsin-Madison Madison, Wl
INTRODUCTIONThe Supervisory Control Alertness Monitoring and Performance Indicators (SCAMPI) project was
begun at the University of Wisconsin-Madison (UW) in early 1995. SCAMPI was developed in the context
of a new UW research center bringing researchers from engineering, social, life, and medical sciences
together to cooperate on improving human-system interactions. The UW Center for Human Performance in
Complex Systems (CHPCS) was identified by both government and corporate participants as a unique
opportunity for pursuing novel directions in technology developments and implementations to reduce human
error and accidents in complex technological systems. One project specified by the Advanced Research
Projects Agency (ARPA: now DARPA) was a need to create "adaptive automation" technologies which were
capable of predicting and counteracting deficits in operator alertness while performing computer-based
supervisory control tasks. Equipment and performance monitoring in adaptive automation is believed to be
able to allow fewer human operators to interact more effectively with complex system processes.I was originally involved in SCAMPI development efforts on the basis of three relevant links to
research conducted in my human factors group performance laboratory. The first link was based on my
interests in information flow and methods to improve information presentation in human-machine interfaces
(HMI) for cooperative task performance. The second link involved my development of quantitative
approaches to analyzing human responses to changing cognitive task demands. These approaches are based
on feedback control engineering systems tools using second-order differential equations of system responses
to changing input functions. The third link incorporated research being conducted by a group at the Naval
Research and Development (NRaD) Center in San Diego (including my graduate student Steven Murray),
looking at human supervisory control of multiple autonomous robots, and real-time neural network
algorithms for identifying operator state during task performance. By spring 1995, the SCAMPI project was
tasked to demonstrate feasibility in three technology development areas:
|• ||Ambulatory, lightweight technologies to collect physiological data (including EEG and
EKG) to be transmitted in "noisy" radio frequency environments;|
|• ||Analysis tools to identify and predict short-term (minute scale) deficits in operator
alertness and associated human performance decrements:|
|• ||Strategies to incorporate negative feedback signals to improve the quality and reliability
of human supervisory control HMI and adaptive automation.|
Technologies to support wearable computing capabilities and ambulatory physiology data collection
have already been in existence for several years, and continue to improve in performance and comfort ( Bass, 1998). Therefore, SCAMPI was not required to conduct specific technology development efforts in
physiological data collection hardware. Similarly, neural science and physiological psychology researchers
have focused on collecting data from specific brain and other sites that are thought to relate to human
cognitive performance tasks ( Makeig,
Posal, 1993; Wilson,
Davis, 1994). Thus,
my tasks as leader of the SCAMPI project were to simply select candidate technologies capable of non-
Questia, a part of Gale, Cengage Learning. www.questia.com
Book title: Automation Technology and Human Performance:Current Research and Trends.
Contributors: Mark W. Scerbo - Editor.
Publisher: Lawrence Erlbaum Associates.
Place of publication: Mahwah, NJ.
Publication year: 1999.
Page number: 213.
This material is protected by copyright and, with the exception of fair use, may
not be further copied, distributed or transmitted in any form or by any means.