Papers
6
Total Citations
79
H-Index
5
About
Kaizhu Huang is a leading researcher in robotics and artificial intelligence, with a primary focus on reinforcement learning, continual learning, and visual perception for autonomous systems. His most impactful work centers on overcoming catastrophic forgetting in sequential multi-task learning—a critical challenge where AI systems lose previously acquired knowledge when learning new tasks. Huang pioneered state primitive learning and encoding primitives generation policy learning, enabling robotic arms to master multiple manipulation tasks sequentially without memory degradation. His research on guided policy search (GPS) has advanced reinforcement learning by addressing local optima and real-time sample collection dilemmas, earning 44 citations. In visual odometry, Huang developed GSL-VO, a lightweight framework that integrates geometric and semantic information to enhance robot perception in dynamic environments—a breakthrough for indoor robotics with 10 citations. His contributions to neural information processing further solidify his expertise. With a publication record spanning top venues and a focus on practical robotic applications, Huang’s work is essential reading for researchers tackling lifelong learning, policy optimization, and vision-based navigation in real-world robotics.
Research Focus
Key Achievements
Top Papers
- 1Guided Policy Search for Sequential Multitask Learning44 citations · 2018
- 2
- 3Neural Information Processing9 citations · 2014
- 4State Primitive Learning to Overcome Catastrophic Forgetting in Robotics7 citations · 2020
- 5
- 6