Utonia: Toward One Encoder for All Point Clouds
This work addresses the problem of developing a unified model for point clouds from various domains, which is significant for applications in AR/VR, robotics, and autonomous driving.
The authors tackled the problem of training a single self-supervised point transformer encoder across diverse domains, achieving a consistent representation space that transfers across domains and improves perception capability. This unification also benefits embodied and multimodal reasoning, with observed gains in robotic manipulation and spatial reasoning.
We dream of a future where point clouds from all domains can come together to shape a single model that benefits them all. Toward this goal, we present Utonia, a first step toward training a single self-supervised point transformer encoder across diverse domains, spanning remote sensing, outdoor LiDAR, indoor RGB-D sequences, object-centric CAD models, and point clouds lifted from RGB-only videos. Despite their distinct sensing geometries, densities, and priors, Utonia learns a consistent representation space that transfers across domains. This unification improves perception capability while revealing intriguing emergent behaviors that arise only when domains are trained jointly. Beyond perception, we observe that Utonia representations can also benefit embodied and multimodal reasoning: conditioning vision-language-action policies on Utonia features improves robotic manipulation, and integrating them into vision-language models yields gains on spatial reasoning. We hope Utonia can serve as a step toward foundation models for sparse 3D data, and support downstream applications in AR/VR, robotics, and autonomous driving.