Rule-VLN: Bridging Perception and Compliance via Semantic Reasoning and Geometric Rectification

TL;DR

Rule-VLN introduces a large-scale urban benchmark for rule-compliant navigation and proposes SNRM, a module that enhances safety awareness in pre-trained embodied AI agents, addressing semantic and regulatory constraints.

Contribution

The paper presents the first large-scale urban benchmark for rule-compliant navigation and a universal zero-shot safety module, SNRM, to improve regulatory adherence in embodied AI.

Findings

Rule-VLN challenges current models with diverse regulatory constraints.

SNRM significantly reduces goal violation rate by 19.26%.

Navigation success rate improves by 5.97% with SNRM.

Abstract

As embodied AI transitions to real-world deployment, the success of the Vision-and-Language Navigation (VLN) task tends to evolve from mere reachability to social compliance. However, current agents suffer from a "goal-driven trap", prioritizing physical geometry ("can I go?") over semantic rules ("may I go?"), frequently overlooking subtle regulatory constraints. To bridge this gap, we establish Rule-VLN, the first large-scale urban benchmark for rule-compliant navigation. Spanning a massive 29k-node environment, it injects 177 diverse regulatory categories into 8k constrained nodes across four curriculum levels, challenging agents with fine-grained visual and behavioral constraints. We further propose the Semantic Navigation Rectification Module (SNRM), a universal, zero-shot module designed to equip pre-trained agents with safety awareness. SNRM integrates a coarse-to-fine visual perception VLM framework with an epistemic mental map for dynamic detour planning. Experiments demonstrate that while Rule-VLN challenges state-of-the-art models, SNRM significantly restores navigation capabilities, reducing CVR by 19.26% and boosting TC by 5.97%.