Weiye Zhao

Papers

Total Citations

H-Index

About

Weiye Zhao is a researcher specializing in safe reinforcement learning, constrained policy optimization, and formal safety verification for autonomous and control systems. His work addresses one of the most critical challenges in deploying reinforcement learning to real-world applications: ensuring that agents reliably satisfy safety constraints throughout their operation, not merely on average. Zhao's most prominent contribution is his comprehensive survey on state-wise safe reinforcement learning (2023), which has accumulated 41 citations and serves as a foundational reference for researchers navigating the rapidly evolving landscape of safety in RL. This work systematically categorizes approaches to enforcing state-wise constraints — among the most practically relevant safety requirements in robotics, autonomous driving, and industrial control. Complementing this, his paper on State-wise Constrained Policy Optimization proposes concrete algorithmic solutions to this challenge, while his earlier work on Safety Index Synthesis via Sum-of-Squares Programming bridges control theory and machine learning by offering principled, automated methods for designing safety index functions under control limitations. Collectively, Zhao's research reflects a rigorous, interdisciplinary approach that connects theoretical guarantees with practical deployment needs, making him an important voice in the growing field of trustworthy and reliable autonomous systems.