Abstract Interpretation for Generalized Heuristic Search in Model-Based Planning

TL;DR

This paper proposes using abstract interpretation as a unifying framework for domain-general model-based planning heuristics, enabling richer models, integration with learning, and broader applicability in complex, uncertain environments.

Contribution

It introduces abstract interpretation as a novel approach to develop and unify heuristics for complex, uncertain, and probabilistic world models in planning.

Findings

Abstract interpretation extends heuristic search to richer models.

Heuristics can incorporate complex data types and uncertainty.

Framework supports integration with learning for improved planning.

Abstract

Domain-general model-based planners often derive their generality by constructing search heuristics through the relaxation or abstraction of symbolic world models. We illustrate how abstract interpretation can serve as a unifying framework for these abstraction-based heuristics, extending the reach of heuristic search to richer world models that make use of more complex datatypes and functions (e.g. sets, geometry), and even models with uncertainty and probabilistic effects. These heuristics can also be integrated with learning, allowing agents to jumpstart planning in novel world models via abstraction-derived information that is later refined by experience. This suggests that abstract interpretation can play a key role in building universal reasoning systems.