Why Not? Solver-Grounded Certificates for Explainable Mission Planning

Researchers developed a new method for explaining satellite mission planning decisions using solver-grounded certificates that directly derive explanations from optimization models. The approach achieves perfect accuracy in explaining why scheduling requests are accepted or rejected, outperforming traditional post-hoc explanation methods that produce non-causal attributions 29% of the time.

• →New faithfulness-first approach generates explanations directly from optimization models rather than independent reasoning layers.
• →Method achieves perfect soundness, counterfactual validity, and stability in explaining satellite scheduling decisions.
• →Traditional post-hoc baselines produce non-causal attributions in 29% of cases and miss constraint conjunctions in multi-cause rejections.
• →Scalability analysis confirms practical extraction times for operational batches up to 200 orders and 30 satellites.
• →Certificates provide minimal infeasible subsets for rejections and tight constraints for selections with what-if query capabilities.