Elizabeth D. Murphy and Kent L. Norman
University of Maryland
For over 30 years, the National Aeronautics and Space Administration (NASA) has been placing unmanned spacecraft in low-Earth orbit for various scientific purposes. Over the years, the role of human operators in these systems has evolved from semi-manual to supervisory control (see, e.g., Moray, 1986; Sheridan, 1976, 1997). Although supervisory control distances the operator from the moment-to-moment operations, it still requires that operators be present to respond to alerts from the system. Until very recently, ground monitoring-and-control of these missions has required the support of human operators on a 24-hour- a-day basis. As advances in technology make it possible to build spacecraft that can operate even more autonomously than their predecessors, a continuous human presence in the control room is no longer needed.
With advances in the technology of ground operations, the required monitoring and control can be handled to a large degree by automated systems in a "lights-out" (or "hands-off") unattended context. In the full lights-out concept, operations would be run autonomously, around the clock, seven days a week. In a partial lights-out environment, there might be one daytime shift staffed by human operators/analysts. At the end of that shift, control-room personnel turn the lights off and go home. When a problem develops that is outside the capabilities of either on-board or ground systems, however, a human expert (or team of experts) is called upon, via a paging system, to perform fault diagnosis and resolution.
Little or no attention has been paid to the cognitive issues associated with the full lights-out mode of operation. The primary issue addressed in the present research is that of human performance, i.e., how to bring the human into the full situational context quickly after a period of prolonged detachment from a particular mission. Given that full context can be provided, a plausible view of the future holds that human operators will be neither monitors nor supervisors of autonomous systems, but consultants who are called in on an exception basis to solve problems beyond the scope of the automation. The present research is designed to investigate implications of that hypothesis.
In the unmanned spacecraft domain, high levels of autonomy are already being introduced, both onboard and on the ground (e.g., Abedini, Moriarta, Biroscak, Losik, & Malina, 1995; Aked & Pylyser, 1996). These advanced software capabilities take engineers and analysts beyond supervisory control into a new paradigm, where they are still part of the system but are needed only when the "intelligent" automation calls for human help. They do not need to be physically present at any particular support facility, and they are not tasked with monitoring mission operations.
Autonomous operations depend on the design, development, and continuing operation of software processes both onboard the spacecraft and embedded in the ground operations system. Onboard capabilities may include automatic safeing of the spacecraft in response to unknown conditions and rule-based fault detection and resolution, such as those long in place on the Voyager 1 spacecraft ( Sawyer, 1998). Alternatively, monitoring and fault detection might also be performed within the ground operations system by "surrogate controllers," that is, automated processes that perform functions formerly performed by human ground-control operators ( Truszkowski, 1996). These processes are also known as "software agents" (e.g., Bradshaw, 1997). Agents developed at NASA-Goddard are distinguished from other automated processes