Rethinking the Embodied Gap in Vision-and-Language Navigation: A Holistic Study of Physical and Visual Disparities

2025-07-17

AI Summary

Existing vision-language navigation (VLN) methods rely excessively on idealized simulation environments and neglect critical challenges in physical embodied deployment—such as kinematic constraints, perception-actuation misalignment, and environmental disturbances. Method: This paper introduces VL-N-PE, the first physically realistic VLN benchmark tailored for humanoid, quadrupedal, and wheeled robots. We propose a multi-technique navigation framework integrating: (i) a classification-based discrete action policy, (ii) diffusion models for dense waypoint generation, and (iii) a training-free, map-augmented large language model for collaborative path planning. Contributions/Results: Comprehensive experiments identify observation limitations, illumination variations, and motion instability (e.g., falling, collisions) as primary causes of performance degradation. We quantitatively demonstrate, for the first time, the severe locomotion bottlenecks of legged robots in complex real-world scenarios and the profound generalization gap of current VLN models under physical embodiment constraints.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Computer Vision: Language and Vision
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Recent Vision-and-Language Navigation (VLN) advancements are promising, but their idealized assumptions about robot movement and control fail to reflect physically embodied deployment challenges. To bridge this gap, we introduce VLN-PE, a physically realistic VLN platform supporting humanoid, quadruped, and wheeled robots. For the first time, we systematically evaluate several ego-centric VLN methods in physical robotic settings across different technical pipelines, including classification models for single-step discrete action prediction, a diffusion model for dense waypoint prediction, and a train-free, map-based large language model (LLM) integrated with path planning. Our results reveal significant performance degradation due to limited robot observation space, environmental lighting variations, and physical challenges like collisions and falls. This also exposes locomotion constraints for legged robots in complex environments. VLN-PE is highly extensible, allowing seamless integration of new scenes beyond MP3D, thereby enabling more comprehensive VLN evaluation. Despite the weak generalization of current models in physical deployment, VLN-PE provides a new pathway for improving cross-embodiment's overall adaptability. We hope our findings and tools inspire the community to rethink VLN limitations and advance robust, practical VLN models. The code is available at https://crystalsixone.github.io/vln_pe.github.io/.

Problem

• Bridging physical and visual disparities in Vision-and-Language Navigation
• Evaluating VLN methods in realistic robotic settings with diverse embodiments
• Addressing performance degradation due to environmental and locomotion constraints

Innovation

• VLN-PE platform supports humanoid, quadruped, wheeled robots
• Diffusion model for dense waypoint prediction
• Map-based LLM integrated with path planning