Applying constraint propagation technology to an expert system planning Hubble Space Telescope observations

A relatively simple yet powerful way to represent temporal constraints is as suitability functions, which identify intervals during which a constraint allows or precludes a given activity. The authors describe the successful application of this framework to Transformation, an expert system for planning Hubble Space Telescope (HST) observations. TRANS is charged with many tasks requiring constraint checking, including ordering observations, grouping those performed together and determining timing links between them. These tasks have been successfully performed using the Micro-Spike package, which supports the generation, propagation and path consistency of temporal constraints.<>