Workshop
2026 Hamiltonian-Guided Diffusion Fields for Variable-Length Rigid-Arm Trajectory Generation
Combines diffusion probabilistic fields with Hamiltonian guidance for variable-length robotic trajectory generation.
University of Illinois Chicago · Computer Science
Guorui Sang
PhD Student & Graduate Research Assistant
I work with Prof. Pedram Rooshenas at UIC. My research focuses on generative models and neural operators for learning from spatial data and generating physically plausible robotic trajectories.
Diffusion ModelsPhysics-Informed MLNeural OperatorsRobotics
Seeking Summer 2027 research internships
Expected PhD graduation: 2030
Selected Publications
Research
Methods I develop and the problems they address
Recursive Function-Space Transformer
Manuscript in preparation
- Proposed Recursive FST for operator learning to preserve local detail and capture long-range dependencies with a compact latent representation.
- Designed input-dependent spatial anchors and recursive attention over a continuous feature field. The model supports queries at arbitrary coordinates.
- Experiments evaluate the framework on PDE prediction and image classification.
Hamiltonian-Guided Trajectory Generation
Extended manuscript in preparation
- Proposed Hamiltonian-guided diffusion to generate physically plausible robotic trajectories across variable horizons without a simulator at inference.
- Formulated an energy function from Hamiltonian residuals for guided sampling. Gradient guidance and importance resampling use this energy to steer the diffusion denoising process.
- Evaluation uses simulated robot-arm trajectories in MuJoCo. This ongoing project extends our ICLR 2026 ReALM-GEN Workshop paper.
ConSinger
ICASSP 2025 · Equal contribution
Jul 2024 - Dec 2024
- Co-developed a consistency model for efficient singing voice synthesis. Designed a quality scorer to select the denoising starting point.
- Reduced inference time by 65% versus DiffSinger on PopCS with comparable perceptual quality (MOS: 3.88 vs. 3.81).
Experience
Graduate Research Assistant
University of Illinois Chicago
Chicago, IL
- Conduct research with Prof. Pedram Rooshenas in diffusion models, physics-informed machine learning, and neural operators.
- Develop and evaluate function-space Transformers and models for robotic trajectory generation using PyTorch.
Graduate Teaching Assistant
University of Illinois Chicago
Chicago, IL
- Led MATLAB labs for more than 30 students, helping students develop practical programming and problem-solving skills.
Education
Ph.D. in Computer Science
University of Illinois Chicago
Chicago, IL, USA
Advisor: Professor Pedram Rooshenas
Expected graduation: 2030. Research in diffusion models, neural operators, and robot dynamics.
M.S. in Computer Science
Beijing University of Technology
Beijing, China
Specialized in conditional image generation using diffusion models.
B.S. in Software Engineering
Hainan University
Haikou, Hainan, China
Skills & Expertise
Languages & Frameworks
PythonC/C++PyTorchPyTorch Lightning
Research Tools
MuJoCoWeights & BiasesGitLinux
Research Areas
Diffusion ModelsNeural OperatorsPhysics-Informed MLRobot Dynamics
Books
Book
2023 《计算机考研精炼 1000 题》（Computer Science Graduate Examination: 1000 Exercises）
Co-author and associate editor under the pen name Yuhang (宇航).
Contributed to the data structures section, organizing core theories and concepts alongside 300 problems and solutions across approximately 170 pages.