REST: Receding Horizon Explorative Steiner Tree for Zero-Shot Object-Goal Navigation

2026-03-19

AI Summary

This work addresses the insufficient subgoal representation in zero-shot object goal navigation, where existing approaches rely solely on isolated waypoints and neglect both the information gain along paths and the structural relationships among candidate trajectories. To overcome this limitation, we propose a training-free hierarchical navigation framework that models the option space as an agent-centric, path-sharing exploratory Steiner tree. Our method constructs an open-vocabulary 3D map from online RGB-D streams and employs a coarse-to-fine large language model reasoning mechanism to textualize candidate paths and select among them via chain-of-thought prompting. Evaluated on Gibson, HM3D, and HSSD benchmarks, our approach achieves state-of-the-art or near-best performance in success rate and demonstrates optimal or near-optimal path efficiency, significantly advancing zero-shot navigation capabilities.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Search and Optimization: Learning to Search
• Computer Vision: Visual Reasoning & Symbolic Representations

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Zero-shot object-goal navigation (ZSON) requires navigating unknown environments to find a target object without task-specific training. Prior hierarchical training-free solutions invest in scene understanding (\textit{belief}) and high-level decision-making (\textit{policy}), yet overlook the design of \textit{option}, i.e., a subgoal candidate proposed from evolving belief and presented to policy for selection. In practice, options are reduced to isolated waypoints scored independently: single destinations hide the value gathered along the journey; an unstructured collection obscures the relationships among candidates. Our insight is that the option space should be a \textit{tree of paths}. Full paths expose en-route information gain that destination-only scoring systematically neglects; a tree of shared segments enables coarse-to-fine LLM reasoning that dismisses or pursues entire branches before examining individual leaves, compressing the combinatorial path space into an efficient hierarchy. We instantiate this insight in \textbf{REST} (Receding Horizon Explorative Steiner Tree), a training-free framework that (1) builds an explicit open-vocabulary 3D map from online RGB-D streams; (2) grows an agent-centric tree of safe and informative paths as the option space via sampling-based planning; and (3) textualizes each branch into a spatial narrative and selects the next-best path through chain-of-thought LLM reasoning. Across the Gibson, HM3D, and HSSD benchmarks, REST consistently ranks among the top methods in success rate while achieving the best or second-best path efficiency, demonstrating a favorable efficiency-success balance.

Problem

• zero-shot object-goal navigation
• option design
• subgoal selection
• path planning
• scene understanding

Innovation

• Zero-shot Object-Goal Navigation
• Option Space as Path Tree
• Training-Free Navigation
• LLM-based Hierarchical Reasoning
• Sampling-based Explorative Planning