MomaGraph: State-Aware Unified Scene Graphs with Vision-Language Models for Embodied Task Planning

Published in International Conference on Learning Representations (ICLR), 2026, 2025