Publications

Publications and ongoing research manuscripts.

2026

  1. ProDrive: Proactive Planning for Autonomous Driving via Ego-Environment Co-Evolution
    ProDrive: Proactive Planning for Autonomous Driving via Ego-Environment Co-Evolution
    Chuyao Fu, Shengzhe Gan, Zhuoli Ouyang, and 6 more authors
    In CVPR GigaBrain Challenge Workshop, 2026
    arXiv
  2. Learning Action-Video Compatibility as a Reference-Free Verifier for Embodied World Models
    Learning Action-Video Compatibility as a Reference-Free Verifier for Embodied World Models
    Chuyao Fu, Xiaojie Zhang, Xiaowei Chi, and 6 more authors
    Under review · 2026
    PDF
  3. Token-World: Policy-Oriented World Modeling in VLM Token Space for Robot Manipulation
    Token-World: Policy-Oriented World Modeling in VLM Token Space for Robot Manipulation
    Chuyao Fu, Xiaowei Chi, Yuhan Rui, and 14 more authors
    Submitted to ICRA 2027
    PDF
  4. AxisField-GS: Feed-Forward Articulation Recovery from Single-State Semantic Gaussian Assets
    AxisField-GS: Feed-Forward Articulation Recovery from Single-State Semantic Gaussian Assets
    Shengzhe Gan*, Chuyao Fu*, Yugong Wang, and 2 more authors
    Submitted to ICASSP 2027. * Equal contribution.
    PDF
  5. Mask World Model: Predicting What Matters for Robust Robot Policy Learning
    Mask World Model: Predicting What Matters for Robust Robot Policy Learning
    Yunfan Lou, Xiaowei Chi, Xiaojie Zhang, and 9 more authors
    ICML, 2026
    arXiv
  6. EchoArena: Mutual Benchmarking of World Models and VLA Policies
    EchoArena: Mutual Benchmarking of World Models and VLA Policies
    Yu-Kai Wang, Kevin Zhang, Xiaowei Chi, and 14 more authors
    In CVPR GigaBrain Challenge Workshop, 2026
    PDF
  7. FORCE: Efficient VLA Reinforcement Fine-Tuning via Value-Calibrated Warm-up and Self-Distillation
    FORCE: Efficient VLA Reinforcement Fine-Tuning via Value-Calibrated Warm-up and Self-Distillation
    Shuyi Zhang, Yunfan Lou, Hongyang Cheng, and 8 more authors
    arXiv preprint, 2026
    arXiv
  8. LJM: A Latent Joint-conditional Model for Action-Conditioned Embodied World Model
    LJM: A Latent Joint-conditional Model for Action-Conditioned Embodied World Model
    Kuangzhi Ge, Yu-Kai Wang, Yuhan Rui, and 13 more authors
    Under review · 2026
    PDF
  9. MirrorSentinel: Suppressing Mirror-Induced Ghost Structures in Elevator LiDAR Maps with Visual Depth
    MirrorSentinel: Suppressing Mirror-Induced Ghost Structures in Elevator LiDAR Maps with Visual Depth
    Yugong Wang, Shengzhe Gan, Yu Zhan, and 3 more authors
    Under review at ROBIO 2026 · 2026
    PDF