arXiv:2512.09903v1 Announce Type: cross 
Abstract: Visual navigation has emerged as a practical alternative to traditional robotic navigation pipelines that rely on detailed mapping and path planning. However, constructing and maintaining 3D maps is often computationally expensive and memory-intensive. We address the problem of visual navigation when exploration videos of a large environment are available. The videos serve as a visual reference, allowing a robot to retrace the explored trajectories without relying on metric maps. Our proposed method, YOPO-Nav (You Only Pass Once), encodes an environment into a compact spatial representation composed of interconnected local 3D Gaussian Splatting (3DGS) models. During navigation, the framework aligns the robot's current visual observation with this representation and predicts actions that guide it back toward the demonstrated trajectory. YOPO-Nav employs a hierarchical design: a visual place recognition (VPR) module provides coarse localization, while the local 3DGS models refine the goal and intermediate poses to generate control actions. To evaluate our approach, we introduce the YOPO-Campus dataset, comprising 4 hours of egocentric video and robot controller inputs from over 6 km of human-teleoperated robot trajectories. We benchmark recent visual navigation methods on trajectories from YOPO-Campus using a Clearpath Jackal robot. Experimental results show YOPO-Nav provides excellent performance in image-goal navigation for real-world scenes on a physical robot. The dataset and code will be made publicly available for visual navigation and scene representation research.

تم تقديم YOPO-Nav كطريقة جديدة للملاحة البصرية تستخدم رسومات Gaussian Splatting ثلاثية الأبعاد (3DGS) المستمدة من مقاطع الفيديو الاستكشافية ذات المرور الواحد، مما يسمح للروبوتات بإعادة تتبع المسارات دون الحاجة إلى خرائط مقياسية مفصلة. يقلل هذا النهج بشكل كبير من المتطلبات الحاسوبية وذاكرة الوصول العشوائي المرتبطة عادةً بأنظمة الملاحة الروبوتية التقليدية.

YOPO-Nav se ha presentado como un nuevo método de navegación visual que utiliza gráficos de 3D Gaussian Splatting (3DGS) derivados de videos de exploración de un solo paso, permitiendo a los robots retrazar caminos sin necesidad de mapas métricos detallados. Este enfoque reduce significativamente las demandas computacionales y de memoria típicamente asociadas con los sistemas de navegación robótica tradicionales.

YOPO-Nav a été introduit comme une nouvelle méthode de navigation visuelle qui utilise des graphes de 3D Gaussian Splatting (3DGS) dérivés de vidéos d'exploration en un seul passage, permettant aux robots de retracer des chemins sans avoir besoin de cartes métriques détaillées. Cette approche réduit considérablement les exigences computationnelles et de mémoire généralement associées aux systèmes de navigation robotique traditionnels.

YOPO-Nav has been introduced as a novel visual navigation method that utilizes 3D Gaussian Splatting (3DGS) graphs derived from one-pass exploration videos, allowing robots to retrace paths without the need for detailed metric maps. This approach significantly reduces the computational and memory demands typically associated with traditional robotic navigation systems.

YOPO-Nav: Visual Navigation using 3DGS Graphs from One-Pass Videos

Was this article worth reading? Share it

CreativeDevJobs

Mapfit

Trupeer.ai

3YOURMIND

Zemith-3bda3b

AIPortalX

Ready to build your own newsroom?