Hierarchical Task Network Planning with LLM-Generated Heuristics

TL;DR

This paper explores using large language models to generate heuristics for hierarchical task network planning, improving search efficiency and coverage compared to traditional methods.

Contribution

It extends LLM-based heuristic generation from classical to hierarchical planning, demonstrating significant efficiency gains in HTN planning benchmarks.

Findings

LLM-generated heuristics nearly match the coverage of top HTN planners.

They reduce search effort on 83% of shared benchmark problems.

Performance is evaluated across six standard HTN domains.

Abstract

HTN planning is a variation of classical planning where, instead of searching for a linear sequence of actions, an algorithm decomposes higher-level tasks using a method library until only executable actions remain. On one hand, this allows one to introduce domain knowledge that can speed up the search for a solution through the method library. On the other hand, it creates challenges that go beyond those of classical state-space search. While recent research produced a number of heuristics and novel algorithms that speed up HTN planning, these heuristics are not yet as informative as those available in classical planning algorithms. We investigate whether large language models (LLMs) can generate effective search heuristics for HTN planning, extending the methodology of Corr\^ea, Pereira, and Seipp (2025) from classical to hierarchical planning. Using the Pytrich planner on six standard total-order HTN benchmark domains, we evaluate heuristics generated by nine LLMs under domain-specific prompting and compare them against the TDG and LMCount domain-independent baselines and the PANDA planner. Our results show that LLM-generated heuristics nearly match the coverage of the best available HTN planner, while substantially reducing search effort on 83% of shared problems.