Xizheng Pang
Papers
3
Total Citations
10
H-Index
2
About
Xizheng Pang is an emerging robotics researcher specializing in autonomous navigation and safe reinforcement learning, with a particular focus on solving real-world challenges in indoor robot navigation. His work addresses one of the field's most pressing problems: ensuring that deep reinforcement learning (DRL)-based navigation systems operate safely and reliably without pre-built maps. Pang's most notable contribution is his development of deep safe reinforcement learning approaches for mapless navigation, introducing constrained optimization algorithms to guarantee safety in end-to-end learning systems — work that has garnered 5 citations since its 2021 publication. Recognizing that purely data-driven approaches struggle in complex indoor environments with "local minima areas," he pioneered novel input encoding methods that integrate local obstacle maps with reinforcement learning, improving navigation robustness in confined spaces. His most recent work extends these ideas into hierarchical reinforcement learning frameworks, incorporating congestion estimation to further enhance navigation through challenging environments. Across his growing body of research, Pang consistently bridges theoretical safe RL methodology with practical robotic deployment challenges. Though early in his career with a total of 10 citations, his focused and progressive contributions position him as a promising voice in the autonomous robot navigation community.
Research Focus
Key Achievements
Top Papers
- 1A Deep Safe Reinforcement Learning Approach for Mapless Navigation5 citations · 2021
- 2
- 3