Hi there! I'm a third-year undergraduate student in Computer Science at School of Electronics and Computer Science, University of Southampton. My research interests lie at the intersection of robotics and control. My previous research has primarily focused on learning-based control within the field of control theory. Beyond theory, I am also interested in real-world robotic systems, particularly embodied intelligence. I was very fortunate to be an undergraduate researcher at the MURO Lab at UC San Diego, under the supervision of Professor Jorge Cortés. I am also very fortunate to be working on my dissertation under the guidance of Professor Zhiwu Huang. Currently, I am also a research assistant at the PAIRS Lab at HKUST (Guangzhou), working with Professor Fangqiang Ding.
Research Projects
Humanoid Imitation of Facial Expressions through Deep Learning (Dissertation)
Supervisor: Prof. Zhiwu Huang · [project website]
Humanoid robots are becoming an increasingly prominent research focus in the field of robotics. In scenarios where humans and robots coexist, such as education, healthcare, and service environments, the ability of humanoid robots to express emotions is playing a vital role. Facial expressions, in particular, not only influence first impressions but also directly affect the naturalness and emotional quality of communication. This project builds upon the X2C dataset and focuses on improving the robot's ability to understand and imitate human facial expressions. We are building deep learning architectures with stronger representational capacity to enhance the accuracy of mapping expressions to robot control commands. At the same time, we are developing a fine-grained emotional labeling scheme that captures expression intensity, blended emotions, and micro-expressions, in order to provide higher-quality supervision signals. These efforts aim to support the generation of facial behaviors that are both more expressive and emotionally nuanced.
Neural Control Lyapunov-Barrier Function Synthesis and Verification (paper in preparation)
We present a framework for jointly synthesizing and formally verifying neural Control Lyapunov-Barrier Function (CLF-CBF) certificates for nonlinear control-affine systems. A neural CLF-CBF pair is first trained to encode stability and safety, then verified across the entire state space by computing provable bounds on a compatibility margin—quantifying whether both conditions can be simultaneously satisfied under bounded control—using CROWN-based convex relaxation. The two stages are unified through a counterexample-guided loop: regions where verification fails are extracted and used to retrain the certificates until formal guarantees are achieved. An adaptive state-space partitioner with gradient-sensitive splitting keeps the verification tractable, and composite barrier functions enable obstacle-rich environments.
Scholarships and Grants
- Nominated for Best Final Year Project (Part III Individual Project), Excellence category, School of Electronics and Computer Science, University of Southampton, Summer 2026
- Selected into ENGAGE Scholarship Program, School of Electronics and Computer Science, University of Southampton, ÂŁ5,000, Summer 2024
- Turing Scheme, for ISRP (International Summer Research Program), provided by UK government, ÂŁ841, Summer 2024
Education
BSc Computer Science (with Honours)
Visiting Undergraduate Researcher
MURO Lab, Department of Mechanical and Aerospace Engineering, UC San Diego
Hobbies
I enjoy playing the piano (amateur Grade 10) and table tennis. I also enjoy reading and photography. Some of my photos are in the gallery page.