Constraint based Plan Distance Metric for Operator-Centric Replanning in Multi-Agent Progression Missions
Year: 2026
Type: conference-paper
Abstract: Multi-agent mission planning systems operating in dynamic environments must frequently adapt their plans as objectives shift and operational conditions evolve. Yet this essential adaptability creates an unexpected challenge: replanning algorithms often generate radically different solutions even when the underlying situation has changed minimally. Human operators, responsible for validating and approving these revised plans, face significant cognitive burden when attempting to reconcile entirely new plan structures with their understanding of what actually changed in the mission context. The core problem transcends simple plan comparison; it requires distinguishing operationally significant modifications from algorithmic artifacts like reordered tasks, reassigned agents, or shifted timings that represent the same fundamental mission strategy. We propose a dual contribution organised around the operator-centric replanning problem. First, we introduce a plan distance metric that is directly embedded into a constraint programming formulation, so that the solver itself is guided toward plans that remain structurally close to their predecessors while optimizing mission objectives. Second, we frame this metric within a mixed-initiative interaction architecture that places the operator in the loop of successive replanning episodes. Unlike post-processing approaches that filter previously found solutions, our method guides the search process itself toward similar plans, yielding replanning solutions that operators can explain and verify against the previous plan.
Source: HAL
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7215621933
