Learning Domain-Independent Heuristics for Grounded and Lifted Planning

Abstract

We present three novel graph representations of planning tasks suitable for learning domain-independent heuristics using Graph Neural Networks (GNNs) to guide search. In particular, to mitigate the issues caused by large grounded GNNs we present the first method for learning domain-independent heuristics with only the lifted representation of a planning task. We also provide a theoretical analysis of the expressiveness of our models, showing that some are more powerful than STRIPS-HGN, the only other existing model for learning domain-independent heuristics. Our experiments show that our heuristics generalise to much larger problems than those in the training set, vastly surpassing STRIPS-HGN heuristics.

Extended version of AAAI 2024 paper

References in corpus (5)

• Distance Encoding: Design Provably More Powerful Neural Networks for Graph Representation Learning
• Graph Neural Networks with Learnable Structural and Positional Representations
• Equivariant and Stable Positional Encoding for More Powerful Graph Neural Networks
• Learning Domain-Independent Planning Heuristics with Hypergraph Networks
• Learning Generalized Policies Without Supervision Using GNNs