Inspire or Predict? Exploring New Paradigms in Assisting Classical Planners with Large Language Models

2025-08-15

AI Summary

To address the challenge of balancing state-space explosion and domain-knowledge effectiveness in large-scale planning, this paper proposes an LLM-augmented hierarchical planning framework. The method decomposes planning problems hierarchically while tightly integrating large language models (LLMs) into the planning process. Its key contributions are two novel LLM-driven paradigms: (1) LLM4Inspire, which generates heuristic guidance to prune the search space; and (2) LLM4Predict, which injects domain knowledge to predict intermediate state conditions—explicitly modeling domain constraints to enhance plan feasibility and computational efficiency. Experimental results across multiple planning domains demonstrate that the framework substantially reduces search-space size. Notably, LLM4Predict achieves faster convergence to feasible solutions than LLM4Inspire, and the overall framework significantly improves planning efficiency without compromising solution quality.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Multiagent Systems: Multiagent Planning

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• 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

Abstract

Addressing large-scale planning problems has become one of the central challenges in the planning community, deriving from the state-space explosion caused by growing objects and actions. Recently, researchers have explored the effectiveness of leveraging Large Language Models (LLMs) to generate helpful actions and states to prune the search space. However, prior works have largely overlooked integrating LLMs with domain-specific knowledge to ensure valid plans. In this paper, we propose a novel LLM-assisted planner integrated with problem decomposition, which first decomposes large planning problems into multiple simpler sub-tasks. Then we explore two novel paradigms to utilize LLMs, i.e., LLM4Inspire and LLM4Predict, to assist problem decomposition, where LLM4Inspire provides heuristic guidance according to general knowledge and LLM4Predict employs domain-specific knowledge to infer intermediate conditions. We empirically validate the effectiveness of our planner across multiple domains, demonstrating the ability of search space partition when solving large-scale planning problems. The experimental results show that LLMs effectively locate feasible solutions when pruning the search space, where infusing domain-specific knowledge into LLMs, i.e., LLM4Predict, holds particular promise compared with LLM4Inspire, which offers general knowledge within LLMs.

Problem

• Addressing state-space explosion in large-scale planning problems
• Integrating LLMs with domain-specific knowledge for valid plans
• Exploring LLM paradigms for problem decomposition and search pruning

Innovation

• LLM-assisted planner with problem decomposition
• LLM4Inspire provides general heuristic guidance
• LLM4Predict infers domain-specific intermediate conditions