Setwise Preferences: A Case Study in High-Resolution EO Satellite Scheduling

In Operations Research and specifically for the high-resolution Earth Observation Satellite (EOS) scheduling problem, setwise preferences introduce both a robustness to the automation of decisions, but also a complexity that significantly expands the preference space making the elicitation and utilization of these preferences a formidable challenge. In EOS scheduling, which is crucial for disaster management, sustainability, agriculture, and security, large areas of interest are often subdivided into smaller image strips, where the value of obtaining adjacent image strips in close temporal proximity is significantly increased due to temporal dependencies between images, and consequently the value of obtaining a particular set is of higher value than just the individual images. This study introduces a novel representative approach designed to streamline the elicitation and integration process of setwise preferences by interpolating outranking results from a sampling technique. The adoption of this approach addresses the challenges posed by the large preference space of setwise preferences, offering a pragmatic approach to incorporating these critical considerations into fully automated decision-making systems. Consequently, this study contributes to the broader field of OR by providing a viable solution to the intricate problem of setwise preference elicitation and integration.