HiPlan: Hierarchical Planning for LLM-Based Agents with Adaptive Global-Local Guidance

TL;DR

HiPlan introduces a hierarchical planning framework for LLM-based agents that enhances decision-making in complex tasks by decomposing goals into milestones and providing adaptive guidance, significantly improving performance.

Contribution

This work presents HiPlan, a novel hierarchical planning method with adaptive global-local guidance and a milestone library, addressing limitations in LLM decision-making for complex, long-horizon tasks.

Findings

Outperforms strong baselines on two benchmarks

Hierarchical components provide complementary benefits

Effective in complex, long-horizon planning scenarios

Abstract

Large language model (LLM)-based agents have demonstrated remarkable capabilities in decision-making tasks, but struggle significantly with complex, long-horizon planning scenarios. This arises from their lack of macroscopic guidance, causing disorientation and failures in complex tasks, as well as insufficient continuous oversight during execution, rendering them unresponsive to environmental changes and prone to deviations. To tackle these challenges, we introduce HiPlan, a hierarchical planning framework that provides adaptive global-local guidance to boost LLM-based agents'decision-making. HiPlan decomposes complex tasks into milestone action guides for general direction and step-wise hints for detailed actions. During the offline phase, we construct a milestone library from expert demonstrations, enabling structured experience reuse by retrieving semantically similar tasks and milestones. In the execution phase, trajectory segments from past milestones are dynamically adapted to generate step-wise hints that align current observations with the milestone objectives, bridging gaps and correcting deviations. Extensive experiments across two challenging benchmarks demonstrate that HiPlan substantially outperforms strong baselines, and ablation studies validate the complementary benefits of its hierarchical components.