A Roadmap to Guide the Integration of LLMs in Hierarchical Planning
This work addresses the problem of enhancing planning performance in AI by integrating LLMs into hierarchical planning, but it is incremental as it provides a preliminary roadmap and baseline rather than a novel solution.
The paper tackles the unexplored integration of Large Language Models (LLMs) into Hierarchical Planning (HP) by proposing a roadmap, including a taxonomy of methods and a benchmark dataset, with initial results showing limited performance from an LLM planner (3% correct plans).
Recent advances in Large Language Models (LLMs) are fostering their integration into several reasoning-related fields, including Automated Planning (AP). However, their integration into Hierarchical Planning (HP), a subfield of AP that leverages hierarchical knowledge to enhance planning performance, remains largely unexplored. In this preliminary work, we propose a roadmap to address this gap and harness the potential of LLMs for HP. To this end, we present a taxonomy of integration methods, exploring how LLMs can be utilized within the HP life cycle. Additionally, we provide a benchmark with a standardized dataset for evaluating the performance of future LLM-based HP approaches, and present initial results for a state-of-the-art HP planner and LLM planner. As expected, the latter exhibits limited performance (3\% correct plans, and none with a correct hierarchical decomposition) but serves as a valuable baseline for future approaches.