Papers
6
Total Citations
114
H-Index
5
About
Chunsheng Liu is a leading researcher in robot manipulation and autonomous systems, with a core focus on bridging the gap between simulation and reality for complex robotic tasks. His major contributions lie in pixel-level grasp detection, where he developed the Grasp-Aware Network, a high-performance method that adaptively predicts grasp poses at the pixel level—overcoming the limitations of discrete gripper configurations. This work, his most cited with 79 citations, has significantly advanced machine vision-based planar grasping in cluttered scenes. Liu also pioneers non-prehensile manipulation through multi-stage reinforcement learning, enabling robots to handle objects without grasping, and has introduced Hierarchical Diffusion Policy (HDP) for contact-rich manipulation trajectory generation. His research on on-policy, pixel-level grasping across the sim-to-real gap (10 citations) further enhances real-world applicability. Notably, Liu extends his expertise to human-robot interaction and multi-machine cooperative systems, including intelligent fire-fighting robots, demonstrating a commitment to practical, safety-critical applications. With a growing citation record and innovative frameworks like HDP, Liu is shaping the future of dexterous, adaptive robotics.
Research Focus
Key Achievements
Top Papers
- 1
- 2Multi-Stage Reinforcement Learning for Non-Prehensile Manipulation13 citations · 2024
- 3
- 4Human–robot interaction-oriented video understanding of human actions5 citations · 2024
- 5
- 6