Each Space Shuttle mission produces more than 25,000 real-time measurements in NASA's mission control center. Within the mission control center, dozens of computer programs analyze these measurements and present results to mission control personnel. Because these programs support the practice of human-in-the-loop control, they serve primarily to present information to mission controllers. The controller's job is to interpret the displayed information to monitor spacecraft and astronaut performance, taking decisions and control actions when necessary for mission success or crew safety.
A single mission controller clearly cannot monitor all 25,000 real-time measurements. The experience of human space flight has evolved into a practice of several mission control disciplines each monitoring and controlling several thousand measurements. The controllers arrange the measurements by function onto dozens of windowed displays each showing a few hundred measurements. Because of the limited screen size, only a few of the displays is visible at any moment. The problem is: how can we use the statistics gathered from previous Space Shuttle missions to automatically select from a list of candidates the most informative display?
In this paper, we first provide a geometric approach to solving this problem, and then show how this geometric approach can be generalized to take statistical information into consideration.