cv

Visiting Scholar

Department of Electrical and Computer Engineering, National University of Singapore

Advisor: Prof. Xingyu Liu

Research Intern

Tsinghua SAIL Group, Tsinghua University

Advisors: Prof. Jun Zhu, Prof. Hang Su.

• Developed ManiBox, a bounding-box-guided manipulation framework based on a scalable simulation-driven teacher-student paradigm, improving policy and spatial generalization.
• Developed PEAC, a Pre-trained Embodiment-Aware Control algorithm that learns embodiment-aware and task-agnostic priors through online interaction in reward-free environments.

Research Intern

Existential Robotics Lab, UC San Diego

Advisor: Prof. Nikolay A. Atanasov.

• Developed methods to prove Lyapunov stability of uncertain dynamical systems and synthesized stabilizing controllers for control-affine systems under model uncertainty.

B.S. (Honours) (Highest Distinction)

National University of Singapore

Mathematics (Major) & Computer Science (Minor)

• Core Courses: Machine Learning (A+), Introduction to Artificial Intelligence (A), AI Planning and Decision Making (A+), Linear Algebra (A+), Calculus (A), Stochastic Process (A), Ordinary Differential Equations (A), Mathematical Modeling (A+), Matrix Computation (A+)
• Research interests: RL, Embodied AI

B.S. (Incomplete, transferred to NUS)

City University of Hong Kong

Biomedical Sciences

PEAC: Unsupervised Pre-training for Cross-Embodiment Reinforcement Learning

NeurIPS 2024

Project Page: https://yingchengyang.github.io/ceurl

ManiBox: Enhancing Spatial Grasping Generalization via Scalable Simulation Data Generation

arXiv

Project Page: https://thkkk.github.io/manibox

Image Deconvolution Using Wavelet Transform

Designed and implemented a Haar wavelet transform framework for denoising periodic signals contaminated with additive Gaussian noise; compared convolution strategies (zero-padding, valid, circular) and thresholding schemes for reconstruction fidelity.

Simulator-Driven Policy Learning for Robotics

Extended Isaac Lab to support custom RL algorithms (A2C, SAC, TD3, DDPG, PPO) for large-scale benchmarking; created a 'Throw a Dice' manipulation task with Franka arm using hierarchical reward shaping and curriculum learning; applied domain randomization across object type/scale/position to improve robustness and diversity.

Safe Policy Modification via Control Barrier Functions

Investigated safety guarantees in RL by embedding Lyapunov-like control barrier functions into nominal controllers to ensure forward invariance; designed a safeguarding controller with formal proofs and simulations across constrained systems.

Bestätigung zum Stipendium / Scholarship (€2,800)

Heidelberg University

Scholarship award.

Hong Kong Tech Seed Fund (HK$10,000)

City University of Hong Kong

Seed fund award.

Dean’s List AY20/21 Sem 1

City University of Hong Kong

Dean’s List AY21/22 Sem 2

City University of Hong Kong