Researchers have introduced XDen-1K, a novel large-scale multimodal dataset designed to advance physical property inference for embodied AI and robotic manipulation. This dataset features 1,000 real-world objects across 137 categories, including detailed 3D geometric models with part-level annotations and paired real-world biplanar X-ray scans. XDen-1K also provides high-fidelity volumetric density fields reconstructed from these scans, establishing a benchmark for density estimation and enabling X-ray-conditioned volumetric segmentation. The dataset's utility is further demonstrated by its ability to improve robotic manipulation performance through derived center-of-mass priors. AI
IMPACT Establishes a new benchmark for physical property inference, crucial for advancing embodied AI and robotic manipulation capabilities.
RANK_REASON The cluster describes a new academic dataset and benchmark published on arXiv. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
