Classical Planning with LLM-Generated Heuristics: Challenging the State of the Art with Python Code

2025-03-24

AI Summary

Large language models (LLMs) exhibit poor generalization in planning tasks, struggling to produce reliable plans—especially on out-of-distribution or scale-up instances. To address this, we propose LLM-Heuristic: a framework that leverages LLMs to directly generate executable and evaluable Python heuristic functions, integrated into a greedy best-first search planner. Unlike prior approaches, it requires no fine-tuning or complex prompt engineering; instead, it samples candidate heuristics from the LLM and automatically validates them via execution-based evaluation to select the optimal one. This is the first work to employ LLMs for end-to-end generation of domain-specific, executable, and empirically assessable heuristic code. Experiments across multiple classical planning domains demonstrate substantial improvements in unseen-task success rates. In several cases, our method expands fewer states than a highly optimized C++ baseline and matches the performance of state-of-the-art learning-based domain-specific planners—achieving both strong generalization and computational efficiency.

Technology Category

• Search and Optimization: Heuristic Search
• Planning, Routing, and Scheduling: Planning with Language Models
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

In recent years, large language models (LLMs) have shown remarkable capabilities in various artificial intelligence problems. However, they fail to plan reliably, even when prompted with a detailed definition of the planning task. Attempts to improve their planning capabilities, such as chain-of-thought prompting, fine-tuning, and explicit "reasoning" still yield incorrect plans and usually fail to generalize to larger tasks. In this paper, we show how to use LLMs to generate correct plans, even for out-of-distribution tasks of increasing size. For a given planning domain, we ask an LLM to generate several domain-dependent heuristic functions in the form of Python code, evaluate them on a set of training tasks within a greedy best-first search, and choose the strongest one. The resulting LLM-generated heuristics solve many more unseen test tasks than state-of-the-art domain-independent heuristics for classical planning. They are even competitive with the strongest learning algorithm for domain-dependent planning. These findings are especially remarkable given that our proof-of-concept implementation is based on an unoptimized Python planner and the baselines all build upon highly optimized C++ code. In some domains, the LLM-generated heuristics expand fewer states than the baselines, revealing that they are not only efficiently computable, but sometimes even more informative than the state-of-the-art heuristics. Overall, our results show that sampling a set of planning heuristic function programs can significantly improve the planning capabilities of LLMs.

Problem

• Improving LLM reliability in generating correct plans
• Developing domain-dependent heuristics via Python code
• Outperforming state-of-the-art domain-independent heuristics

Innovation

• LLM-generated domain-dependent heuristic functions
• Greedy best-first search for heuristic evaluation
• Python code implementation for planning heuristics