Hengqiang Huan
Papers
2
Total Citations
23
H-Index
2
About
Hengqiang Huan is a robotics researcher specializing in reinforcement learning and manipulation, with a focus on bridging the gap between simulation and real-world robotic control. His work centers on two key challenges: enabling robots to perform complex non-prehensile manipulation—moving objects without grasping them—and improving grasp detection in cluttered environments. In his 2024 paper "Multi-Stage Reinforcement Learning for Non-Prehensile Manipulation" (13 citations), Huan introduced a novel multi-stage learning framework that allows robots to combine multiple skills for flexible object manipulation, moving beyond the single-skill limitations of prior methods. His 2023 study "On-Policy and Pixel-Level Grasping Across the Gap Between Simulation and Reality" (10 citations) addresses the persistent sim-to-real transfer problem by developing a method that trains grasp detection directly on pixel-level data rather than 3D models, achieving more robust performance in cluttered scenes. Though early in his career, Huan's work demonstrates significant impact by tackling fundamental bottlenecks in robotic manipulation—combining theoretical advances in reinforcement learning with practical solutions for real-world deployment.
Research Focus
Key Achievements
Top Papers
- 1Multi-Stage Reinforcement Learning for Non-Prehensile Manipulation13 citations · 2024
- 2