Think before Go: Hierarchical Reasoning for Image-goal Navigation

2026-04-19

AI Summary

This work addresses the challenges of sparse visual cues and aimless agent wandering in image-based object navigation, particularly when targets are distant or located across rooms. To tackle these issues, the authors propose HRNav, a novel framework that introduces a human-like hierarchical cognitive mechanism for the first time: a high-level module leverages a vision-language model trained on a self-constructed planning dataset to generate short-horizon navigational paths, while a low-level module executes actions via online reinforcement learning. Crucially, the framework incorporates an innovative Wandering Suppression Penalty (WSP) mechanism to minimize unproductive exploration. Extensive experiments demonstrate that HRNav significantly outperforms existing methods in both simulated and real-world environments, achieving higher success rates in long-range navigation while effectively curbing aimless roaming behavior.

Technology Category

• Humans and AI: Human-Aware Planning and Behavior Prediction
• Planning, Routing, and Scheduling: Planning with Language Models
• Intelligent Robots: Localization, Mapping, and Navigation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work

Abstract

Image-goal navigation steers an agent to a target location specified by an image in unseen environments. Existing methods primarily handle this task by learning an end-to-end navigation policy, which compares the similarities of target and observation images and directly predicts the actions. However, when the target is distant or lies in another room, such methods fail to extract informative visual cues, leading the agent to wander around. Motivated by the human cognitive principle that deliberate, high-level reasoning guides fast, reactive execution in complex tasks, we propose Hierarchical Reasoning Navigation (HRNav), a framework that decomposes image-goal navigation into high-level planning and low-level execution. In high-level planning, a vision-language model is trained on a self-collected dataset to generate a short-horizon plan, such as whether the agent should walk through the door or down the hallway. This downgrades the difficulty of the long-horizon task, making it more amenable to the execution part. In low-level execution, an online reinforcement learning policy is utilized to decide actions conditioned on the short-horizon plan. We also devise a novel Wandering Suppression Penalty (WSP) to further reduce the wandering problem. Together, these components form a hierarchical framework for Image-Goal Navigation. Extensive experiments in both simulation and real-world environments demonstrate the superiority of our method.

Problem

• image-goal navigation
• visual cues
• wandering problem
• long-horizon task
• unseen environments

Innovation

• Hierarchical Reasoning
• Image-goal Navigation
• Vision-Language Model
• Reinforcement Learning
• Wandering Suppression