Learning Domain-Independent Planning Heuristics with Hypergraph Networks

TL;DR

This paper introduces a novel hypergraph neural network framework to learn domain-independent planning heuristics from scratch, capable of generalizing across various problems and domains, and competitive with existing heuristics.

Contribution

It presents the first method to learn planning heuristics directly from hypergraph representations, generalizing across domains and outperforming traditional delete-relaxation heuristics.

Findings

Learned heuristics are competitive with LM-cut.

Heuristics generalize across unseen domains.

Hypergraph networks effectively model delete-relaxation.

Abstract

We present the first approach capable of learning domain-independent planning heuristics entirely from scratch. The heuristics we learn map the hypergraph representation of the delete-relaxation of the planning problem at hand, to a cost estimate that approximates that of the least-cost path from the current state to the goal through the hypergraph. We generalise Graph Networks to obtain a new framework for learning over hypergraphs, which we specialise to learn planning heuristics by training over state/value pairs obtained from optimal cost plans. Our experiments show that the resulting architecture, STRIPS-HGNs, is capable of learning heuristics that are competitive with existing delete-relaxation heuristics including LM-cut. We show that the heuristics we learn are able to generalise across different problems and domains, including to domains that were not seen during training.