LATIDIA · Robótica

GraFT: A Training-Free Framework for Spatial Reasoning in Multimodal Large Language Models via 3D Scene Graphs

arXiv:2609.03892v2 Announce Type: replace-cross Abstract: El razonamiento espacial 3D sustenta la comprensión y la actuación en el mundo físico, pero sigue siendo poco fiable en los actuales modelos multimodales de lenguaje grande (MLLM).

La noticia

arXiv:2609.03892v2 Announce Type: replace-cross Abstract: 3D spatial reasoning underpins understanding and acting in the physical world, yet it remains unreliable in current multimodal large language models (MLLMs). These models falter at precise geometric measurement, at transforming between egocentric and allocentric viewpoints, and at grounding fine-grained appearance. The most common remedies fine-tune the model on large-scale curated spatial-reasoning datasets or attach dedicated encoders for 3D geometry, which typically couples the solution to costly supervision and a specific backbone. We instead introduce GraFT, a training-free framework that supplies the missing 3D structure through a compact, easily maintained 3D scene graph (3DSG). GraFT supplies the graph's 3D structure to the frozen model in