Zeyang Liu
Papers
1
Total Citations
6
H-Index
1
About
Zeyang Liu is a rising researcher at the forefront of robotics and reinforcement learning, with a focused expertise in offline reinforcement learning (RL) for robot manipulation. His major contribution lies in addressing a critical challenge: enabling robots to learn effective policies from static datasets without risky real-time interaction. In his most-cited work, "Improving Offline Reinforcement Learning With in-Sample Advantage Regularization for Robot Manipulation" (2024), Liu introduces a novel regularization technique that stabilizes policy learning by focusing on in-sample data, thereby enhancing both safety and efficiency. This work has already garnered 6 citations, signaling its growing influence in the field. By mitigating the distributional shift problem inherent in offline RL, Liu's approach paves the way for more reliable and sample-efficient robotic systems. His research is particularly impactful for real-world applications where exploration is costly or dangerous, such as industrial automation and assistive robotics. As a young scholar, Liu is establishing himself as a key contributor to the next generation of data-driven, safe robot learning.
Research Focus
Key Achievements
Top Papers
- 1