arXiv:2605.29879v2 Announce Type: replace 
Abstract: Integrating open-vocabulary semantic information into dynamic 3D scene representations is essential for long-term embodied scene understanding. However, existing methods often suffer from fragile instance association due to incomplete cross-view cues, while their limited ability to handle object-level topological changes restricts long-term robotic task execution. Moreover, current 3D scene understanding methods either rely on simple feature matching without explicit spatial reasoning or assume offline ground-truth 3D geometry. To address these challenges, we present DGSG-Mind, a hybrid instance-aware 3D Gaussian dynamic scene graph system with an embodied reasoning agent. Our system couples a probabilistic voxel grid with explicit 3D Gaussians to enable robust cross-modal instance fusion and incremental semantic mapping. It handles dynamic changes through Gaussian-based visual relocalization and localized masked refinement guided by geometric-semantic consistency. Built on the instance Gaussian map, DGSG-Mind further constructs a hierarchical scene graph and develops the 3D Gaussian Mind, which integrates structural relations, spatial-semantic information, and visually annotated RoI Gaussian renderings for multimodal reasoning. Extensive experiments show that DGSG-Mind achieves the best zero-shot 3DVG performance among methods operating on self-reconstructed maps, while also delivering strong performance in 3D open-vocabulary semantic segmentation and scene reconstruction. We further deploy DGSG-Mind on real-world robots to demonstrate its target-oriented reasoning and dynamic update capabilities. The project page of DGSG-Mind is available at https://icr-lab.github.io/DGSG-Mind

يقدم DGSG-Mind نظامًا هجينًا جديدًا لرسم خرائط المشاهد الديناميكية ثلاثية الأبعاد باستخدام Gaussian، بهدف تحسين فهم المشاهد وتثبيتها على المدى الطويل للروبوتات. يدمج هذا النظام معلومات دلالية ذات مفردات مفتوحة ويستخدم شبكة فوكسيول احتمالية مع Gaussian ثلاثي الأبعاد صريح، مما يعالج التحديات المتعلقة بربط الكيانات والتغيرات الطوبولوجية على مستوى الكائنات.

DGSG-Mind presenta un nuevo sistema híbrido de grafo de escena dinámica en 3D que utiliza Gaussianos, con el objetivo de mejorar la comprensión y el anclaje de escenas a largo plazo para la robótica. Este sistema integra información semántica de vocabulario abierto y emplea una cuadrícula de vóxeles probabilística con Gaussianos 3D explícitos, abordando desafíos en la asociación de instancias y cambios topológicos a nivel de objeto.

DGSG-Mind présente un nouveau système hybride de graphes de scène dynamiques en 3D utilisant des Gaussiens, visant à améliorer la compréhension et l'ancrage des scènes à long terme pour la robotique. Ce système intègre des informations sémantiques à vocabulaire ouvert et utilise une grille de voxels probabiliste avec des Gaussiens 3D explicites, répondant aux défis d'association d'instances et de changements topologiques au niveau des objets.

DGSG-Mind introduces a novel hybrid instance-aware 3D Gaussian dynamic scene graph system aimed at enhancing long-term scene understanding and grounding for robotics. This system integrates open-vocabulary semantic information and employs a probabilistic voxel grid with explicit 3D Gaussians, addressing challenges in instance association and object-level topological changes.