Papers
5
Total Citations
16
H-Index
3
About
Yixing Lan is a researcher advancing the frontiers of reinforcement learning (RL) with a focus on sample efficiency, safety, and real-world robotic applications. Their work addresses critical bottlenecks in RL, particularly in continuous control tasks with sparse rewards—a longstanding challenge for deploying RL in robotics. Lan’s key contributions include developing data-efficient algorithms like the least-squares soft Bellman residual method for robotic grasping, which reduces the sample complexity of learning manipulation policies. They have also pioneered safe RL techniques through game-theoretic constrained policy optimization, enabling agents to optimize task performance while satisfying safety constraints—essential for real-world deployment. With over 16 citations across their most-cited papers, Lan’s research has gained traction for tackling practical limitations such as low sample efficiency and local convergence in policy evaluation. Notable achievements include a comprehensive survey on transfer RL, synthesizing progress in knowledge transfer across tasks, and a novel least-squares truncated temporal-difference method for more stable value function estimation. Lan’s work bridges the gap between theoretical RL advances and practical robotic systems, making their research particularly valuable for students and practitioners working on autonomous systems, manipulation, and safe decision-making.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5A Survey on Transfer Reinforcement Learning2 citations · 2025