The paper releases a procedural synthetic dataset of single-cube scenes with ground-truth pose matrices as a foundation for training VLMs in spatial reasoning.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.AI 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Towards Embodied Cognition in Robots via Spatially Grounded Synthetic Worlds
The paper releases a procedural synthetic dataset of single-cube scenes with ground-truth pose matrices as a foundation for training VLMs in spatial reasoning.