Graph-Assisted Stitching for Offline Hierarchical Reinforcement Learning

TL;DR

The paper introduces Graph-Assisted Stitching (GAS), a novel offline hierarchical reinforcement learning framework that uses graph search and state clustering to improve task efficiency and performance across various domains.

Contribution

GAS formulates subgoal selection as a graph search problem using state clustering in TDR space, bypassing high-level policy learning limitations.

Findings

GAS significantly outperforms prior methods in locomotion, navigation, and manipulation tasks.

In the most critical task, GAS achieves a score of 88.3, far exceeding the previous 1.0.

The Temporal Efficiency metric improves graph quality and task success.

Abstract

Existing offline hierarchical reinforcement learning methods rely on high-level policy learning to generate subgoal sequences. However, their efficiency degrades as task horizons increase, and they lack effective strategies for stitching useful state transitions across different trajectories. We propose Graph-Assisted Stitching (GAS), a novel framework that formulates subgoal selection as a graph search problem rather than learning an explicit high-level policy. By embedding states into a Temporal Distance Representation (TDR) space, GAS clusters semantically similar states from different trajectories into unified graph nodes, enabling efficient transition stitching. A shortest-path algorithm is then applied to select subgoal sequences within the graph, while a low-level policy learns to reach the subgoals. To improve graph quality, we introduce the Temporal Efficiency (TE) metric, which filters out noisy or inefficient transition states, significantly enhancing task performance. GAS outperforms prior offline HRL methods across locomotion, navigation, and manipulation tasks. Notably, in the most stitching-critical task, it achieves a score of 88.3, dramatically surpassing the previous state-of-the-art score of 1.0. Our source code is available at: https://github.com/qortmdgh4141/GAS.