Quanyuan Ruan

πŸ“§ ruanquanyuan@outlook.com | πŸ“ž 17820470783 | πŸ™ github | πŸ†Ž δΈ­ζ–‡

Education

Projects

  • Prof. Robot: Differentiable Robot Rendering Without Static and Self-Collisions

    • 2024/07 - 2024/11
    • Researched collision issues in differentiable robot rendering, combining binary classification with SDF fields to achieve collision detection and elimination for robots
    • Used dependent multi-layer perceptrons to learn SDF fields from robot poses to determine collisions, improving gradient optimization reliability
    • Converted SDF fields to collision probabilities, using classification loss to eliminate need for exact manifold distance labels without explicit manifold distance calculations

  • HumanPlus

    • 2024/07 - 2024/09
    • Replicated HumanPlus’s HIT model, implemented collaborative grasping between Rokae robotic arm and Inspire dexterous hand, and optimized grasping strategies
    • Developed DexSim simulation engine to collect simulated environment data, trained models through domain randomization for real-world transfer

  • Vision-Audio-Fusion Perception Robot

    • 2024/03 - 2024/07
    • Used generative multimodal large models to solve open-world visual question answering for robots
    • Integrated language model ChatGLM and vision model CogVLM on robot to improve question answering and visual cognition capabilities
    • Built independent Q&A logic based on FastAPI for flexible adjustment and optimization of Q&A flow

  • C++ Based Ray Tracing Software Rasterizer

    • 2023/12 - 2024/03
    • Implemented C++ based ray tracing software rasterizer capable of rendering high-quality 3D scenes
    • Achieved multi-threading acceleration, anti-aliasing, supported various materials like diffuse, metal, dielectric, and implemented shadows, reflections, refractions

  • Digital Watermarking System

    • 2021/11 - 2021/12
    • Embedded hidden watermark information (invisible) in image spatial, frequency and wavelet domains, enabling digital watermark embedding and extraction
    • Tested watermark robustness against compression, cropping, rotation operations and verified extraction effectiveness

  • Controllable Acoustic Levitation

    • 2020/06 - 2020/09
    • Achieved contactless levitation of lightweight objects using ultrasonic standing waves
    • Adjusted ultrasonic generator amplitude for precise control of levitated objects

Internships

  • The Chinese University of Hong Kong (Shenzhen)

    • 2024/07 - Present
    • Research Assistant (RA) | Shenzhen, Guangdong

  • DexForce (Shenzhen) Intelligent Digital Technology Co., Ltd

    • 2022/12 - 2023/04
    • Software Development Intern | Shenzhen, Guangdong
      • Responsible for algorithm integration in infrastructure, providing technical support for company’s business
      • Developed memory detection algorithms to locate and fix memory leaks in CPU and GPU
      • Became Open3D Contributor, fixed Open3D OBB bug and submitted patch

Publications

  • Quanyuan Ruan, Jiabao Lei, Wenhao Yuan, Yanglin Zhang, Dekun Lu, Guiliang Liu, Kui Jia, Prof. Robot: Differentiable Robot Rendering Without Static and Self-Collisions, CVPR, 2025, CCF-A

Skills

  • Programming: Python, C, C++, CUDA
  • Courses:
    • Math: Probability and Statistics
    • CV: Signals and Systems, Digital Signal Processing, Digital Image Processing, Deep Learning
    • CG: Introduction to Modern Computer Graphics, Geometric Modeling and Processing, Introduction to Physics-Based Computer Animation, High-Quality Real-Time Rendering
  • Containers: Docker
  • Languages: CET4, CET6