🀁 Nan Wang

I build neural rendering and world-model systems at the Beijing Academy of Artificial Intelligence (BAAI), exploring how photorealistic simulation accelerates VR, robotics, and autonomous driving.

I earned my M.Sc. in Computer Science from Tongji University and collaborate closely with teams across research and industry.

Neural Rendering World Models 3D Vision
  • Current Algorithm Engineer · BAAI
  • Focus Simulation · Neural rendering · Robotics
Email Scholar GitHub X LinkedIn
Portrait of Nan Wang

News

    Research

    Representative papers are highlighted.

    HOLO dataset overview

    HOLO: Holistic Lightweight Optimization for Scene Understanding with Auto-Annotation and Multimodal Learning

    Xiaoyun Hu*, Xiaohan Yan*, Nan Wang, Xiaowei Song, Gang Wei, Zhicheng Wang

    WACV, 2026

    We build a large-scale auto-annotated scene-understanding dataset and a lightweight 3D-LLM that efficiently reasons across modalities.

    PUGS physical understanding

    PUGS: Zero-shot Physical Understanding with Gaussian Splatting

    Yinghao Shuai, Ran Yu, Yuantao Chen, Zijian Jiang, Xiaowei Song, Nan Wang, Jv Zheng, Jianzhu Ma, Meng Yang, Zhicheng Wang, Wenbo Ding, Hao Zhao

    ICRA, 2025

    Project Paper Code

    Gaussian splatting reconstructions paired with physical priors enable zero-shot predictions of material and dynamics properties.

    SGGS rendering Alternative SGGS rendering

    Semantic-Guided Gaussian Splatting with Deferred Rendering

    Nan Wang, Xiaohan Yan, Xiaowei Song, Zhicheng Wang

    ICASSP, 2025

    Paper Poster

    Semantic cues from 2D foundation models guide material optimization, yielding expressive deferred rendering for 3DGS.

    GreedyAgent meta-learning

    GreedyAgent: A Simple yet Efficient Approach for Meta-learning from Learning Curves

    Jinyu He, Xiaowei Song, Xiaohan Yan, Nan Wang, Yuqi Miao, Zijian Jiang, Fei Chao, Yan Zhang, Shengchuan Zhang, Rongrong Ji

    ICIC, 2024 (Oral)

    Paper Code

    A greedy meta-learner that leverages learning-curve statistics for fast adaptation across tasks.

    AttenPoint segmentation

    AttenPoint: Exploring Point Cloud Segmentation through Attention-Based Modules

    Xiaohan Yan, Nan Wang, Xiaowei Song, Gang Wei, Zhicheng Wang

    PRCV, 2024

    Paper

    Attention modules blending local and global structure deliver data-efficient point cloud segmentation.

    Projects

    Real2Sim pipeline

    A Real2Sim Pipeline for Robotics Simulation

    Xiaomi · 2024-06

    Code

    3D Gaussian Splatting supplies photoreal rendering while ISAAC Sim handles physics, creating realistic rehearsal spaces for robotics.

    AssistGS project

    3DGS Implementation for Structured-NeRF: Hierarchical Scene Graph with Neural Representation

    DISCOVER Robotics · 2023-11

    Paper Code

    A 3DGS-based replica of Structured-NeRF that decomposes scenes into composable assets for interactive manipulation.

    RandLA and SQN implementations

    PyTorch-Lightning Implementations of RandLA-Net & SQN

    2023-10

    RandLA SQN

    Clean training scripts for large-scale point cloud segmentation under weak supervision.

    LLM Science Exam project

    LLM Science Exam · Using LLMs to Answer Difficult Science Questions

    Kaggle · 2023-10

    Code

    A silver-medal RAG system that ensembles three DeBERTa variants to reason over curated scientific corpora.

    Short Bio

    I am an M.Sc. student in Computer Science at the CAD Research Center, Tongji University, focusing on 3D vision, neural rendering, and world models.

    Misc

    Music 🎶

    Sports 🏃‍♂️

    • Basketball
    • Badminton
    • Swimming
    • Flying Disc

    Languages 💬

    • 中文
    • English
    • 日本語 (learning)
    • Français (learning)

    Others 💡