Hongzhan Yu
PhD Student
Computer Science and Engineering
University of California, San Diego
Office: CSE 4254
Email: hoy021 at ucsd dot edu
I am a Ph.D. student in CSE at UCSD, advised by Professor Sicun Gao.
I earned my M.S. in CSE from UCSD in 2021 and my B.S. in EECS from UCB in 2019.
Research
- I develop safe and resilient learning-based control strategies for robotic systems by integrating data-driven function approximators (e.g. deep neural networks) with formal tools from control theory to obtain provable stability and safety guarantees.
- I investigate versatile, general-purpose robotic controllers grounded in diffusion and flow-matching learning paradigms, and in large-scale foundation models, aiming to endow robots with broad task generalization and rapid adaptation capabilities.
- My long-term objective is to engineer autonomous systems that couple strong theoretical reliability with aggressive performance optimization, enabling trustworthy, high-efficiency automation in real-world deployments.
Papers
- Sequence Modeling for Time-Optimal Quadrotor Trajectory Optimization with Sampling-based Robustness Analysis [arXiv]
Katherine Mao, Hongzhan Yu, Ruipeng Zhang, Igor Spasojevic, M Ani Hsieh, Sicun Gao, and Vijay Kumar
CoRL (Conference on Robot Learning) 2025
- Safe Human Robot Navigation in Warehouse Scenario [arXiv]
Seth Farrell*, Chenghao Li*, Hongzhan Yu, Ryo Yoshimitsu, Sicun Gao, and Henrik I Christensen
CASE (IEEE International Conference on Automation Science and Engineering) 2025
- Estimating Control Barriers from Offline Data [arXiv]
Hongzhan Yu, Seth Farrell, Ryo Yoshimitsu, Zhizhen Qin, Henrik Christensen, and Sicun Gao
ICRA (IEEE International Conference on Robotics and Automation) 2025
- Activation-Descent Regularization for Input Optimization of ReLU Networks [arXiv]
Hongzhan Yu and Sicun Gao
ICML (International Conference on Machine Learning) 2024
- Sequential Neural Barriers for Scalable Dynamic Obstacle Avoidance [project page]
Hongzhan Yu, Chiaki Hirayama, Chenning Yu, Sylvia Herbert, and Sicun Gao
IROS (International Conference on Intelligent Robots and Systems) 2023 [IROS RoboCup Best Paper Award]
- Learning Control Admissibility Models with Graph Neural Networks for Multi-Agent Navigation [project page]
Chenning Yu, Hongzhan Yu, and Sicun Gao
CoRL (Conference on Robot Learning) 2022
Workshop Papers
- Out-of-Distribution-Aware Control Barrier Estimation [link]
Hongzhan Yu and Sicun Gao
Workshop on Out-of-Distribution Generalization in Robotics at RSS, 2025 [Award Nominee]
- Sequence Modeling for Time-Optimal Quadrotor Trajectory Optimization with Sampling-based Robustness Analysis [arXiv]
Katherine Mao, Hongzhan Yu, Ruipeng Zhang, Igor Spasojevic, M Ani Hsieh, Sicun Gao, and Vijay Kumar
Workshop on Leveraging Implicit Methods for Aerial Autonomy at RSS, 2025
Teaching Assistant Experience
- UCSD CSE150: Introduction to AI: Search and Reasoning (Spring 2022)
- UCSD CSE257: Search and Optimization (Winter 2024, Winter 2023, Fall 2021)
Vision without execution is hallucination. -- Thomas Edison