Publications

  • LAC-Net: Linear-Fusion Attention-Guided Convolutional Network for Accurate Robotic Grasping Under the Occlusion, IROS 2024 (Oral Pitch).
    Jinyu Zhang, Yongchong Gu, Jianxiong Gao, Haitao Lin, Qiang Sun, Xinwei Sun, Xiangyang Xue, Yanwei Fu.
    [PDF] [Code] [Project]

    BibTeX 
    @misc{zhang2024lacnetlinearfusionattentionguidedconvolutional,
      title={LAC-Net: Linear-Fusion Attention-Guided Convolutional Network for Accurate Robotic Grasping Under the Occlusion}, 
      author={Jinyu Zhang and Yongchong Gu and Jianxiong Gao and Haitao Lin and Qiang Sun and Xinwei Sun and Xiangyang Xue and Yanwei Fu},
      year={2024},
      eprint={2408.03238},
      archivePrefix={arXiv},
      primaryClass={cs.RO},
      url={https://arxiv.org/abs/2408.03238}, 
}

Open-Source Contributions

  • InternManip — Contributor & Maintainer
    A unified framework for robotic training and evaluation.
    GitHub Repo

  • IROS 2025 Challenge - Manipulation Dataset — Contributor & Maintainer
    Benchmark dataset for the IROS 2025 Challenge Manipulation Track, organized into pretrain, train, and validation splits.
    Hugging Face

  • InternData-M1 — Contributor & Maintainer
    A large-scale embodied robotics dataset (~250k simulated demonstrations) with rich annotations, including 2D/3D boxes, trajectories, grasp points, and semantic masks.
    Hugging Face

  • IROS 2025 Challenge Organization — Contributor & Maintainer
    One of the core contributors of the IROS 2025 Challenge on Dual-Arm Manipulation, featuring 10 carefully designed scenarios with diverse tasks, assessing multi-scale robotic capabilities.