Skip to content

Source paper

Utonia: Toward One Encoder for All Point Clouds

Yujia Zhang, Xiaoyang Wu, Yunhan Yang and 6 more

Yujia Zhang, Xiaoyang Wu, Yunhan Yang, Xianzhe Fan, Han Li, Yuechen Zhang, Zehao Huang, Naiyan Wang, Hengshuang Zhao

arXiv 2603.03283 · 2026126 upvotes on Hugging Face

Research ideas from this paper (3)

Generated by AI. The paper’s authors did not write these.

AI summary

This paper introduces Utonia, a unified point cloud encoder designed to learn a single representation across diverse domains, including indoor scenes, outdoor LiDAR, and object-centric CAD models. By utilizing causal modality blinding to handle missing sensors (color/normals) and perceptual granularity rescaling with RoPE to harmonize varying spatial densities, Utonia achieves strong performance on downstream tasks like segmentation and robotic manipulation using a single shared model.

How it works (AI summary)

The model utilizes a Point Transformer V3 (PTv3) backbone trained via teacher-student self-distillation on a massive mixture of cross-domain datasets. Key techniques include Causal Modality Blinding (randomly dropping input channels to prevent shortcut learning), Perceptual Granularity Rescale (normalizing coordinate scales across domains), and 3D Rotary Positional Embeddings (RoPE) applied to these rescaled coordinates to encode relative geometry robustly.