Linhai Xie
Papers
8
Total Citations
365
H-Index
6
About
Linhai Xie is a robotics researcher whose work sits at the intersection of deep reinforcement learning, autonomous navigation, and non-contact physiological sensing. His most influential contribution, "Towards Monocular Vision based Obstacle Avoidance through Deep Reinforcement Learning" (136 citations), tackles the fundamental challenge of enabling robots to navigate safely using only a single camera—a notoriously difficult problem due to the lack of 3D depth information. To address the inefficiency of training deep RL policies from scratch, Xie introduced "Learning with Training Wheels" (83 citations), a clever framework that uses a simple controller to bootstrap learning, dramatically accelerating training for real-world deployment. He further advanced mapless navigation with "SnapNav" and "Learning with Stochastic Guidance," which together address policy transfer and variance reduction in complex environments. Demonstrating remarkable breadth, Xie also pioneered "Heart Rate Sensing with a Robot Mounted mmWave Radar" (70 citations), enabling mobile robots to monitor vital signs as they move through a home—a breakthrough for elderly care and post-operative monitoring. His work on DEFO-NET extends into physical reasoning, using GANs to predict object deformation from a single image. Across these contributions, Xie consistently pushes toward practical, deployable robotic systems that learn efficiently and operate robustly in the real world.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3Heart Rate Sensing with a Robot Mounted mmWave Radar70 citations · 2020
- 4Learning With Stochastic Guidance for Robot Navigation46 citations · 2020
- 5
- 6
- 7Learning with Stochastic Guidance for Navigation4 citations · 2018
- 8Reinforcement learning based mapless robot navigation3 citations · 2019