Haoyi Zhu

$\pi^3$: Permutation-Equivariant Visual Geometry Learning

Learning Primitive Embodied World Models: Towards Scalable Robotic Learning

OmniWorld: A Multi-Domain and Multi-Modal Dataset for 4D World Modeling

WinT3R: Window-Based Streaming Reconstruction with Camera Token Pool

DeepVerse: 4D Autoregressive Video Generation as a World Model

Bridge Thinking and Acting: Unleashing Physical Potential of VLM with Generalizable Action Expert

CoMo: Learning Continuous Latent Motion from Internet Videos for Scalable Robot Learning

SPA: 3D Spatial-Awareness Enables Effective Embodied Representation

Tra-MoE: Scaling Trajectory Prediction Models for Adaptive Policy Conditioning