Scaling Multi-Agent Epistemic Planning through GNN-Derived Heuristics

TL;DR

This paper introduces GNN-based heuristics to improve the scalability of multi-agent epistemic planning by effectively estimating state quality within graph-structured models, enabling more efficient planning in complex multi-agent scenarios.

Contribution

We develop GNN-derived heuristics tailored for Kripke-structured epistemic states, significantly enhancing the scalability of multi-agent epistemic planning algorithms.

Findings

GNN heuristics outperform baseline methods in scalability.

The approach generalizes across different planning instances.

Planning efficiency improves with GNN-guided search.

Abstract

Multi-agent Epistemic Planning (MEP) is an autonomous planning framework for reasoning about both the physical world and the beliefs of agents, with applications in domains where information flow and awareness among agents are critical. The richness of MEP requires states to be represented as Kripke structures, i.e., directed labeled graphs. This representation limits the applicability of existing heuristics, hindering the scalability of epistemic solvers, which must explore an exponential search space without guidance, resulting often in intractability. To address this, we exploit Graph Neural Networks (GNNs) to learn patterns and relational structures within epistemic states, to guide the planning process. GNNs, which naturally capture the graph-like nature of Kripke models, allow us to derive meaningful estimates of state quality -- e.g., the distance from the nearest goal -- by generalizing knowledge obtained from previously solved planning instances. We integrate these predictive heuristics into an epistemic planning pipeline and evaluate them against standard baselines, showing improvements in the scalability of multi-agent epistemic planning.