Jiangshan Liu
Papers
1
Total Citations
5
H-Index
1
About
Dr. Jiangshan Liu is a rising leader in robotic manipulation, with a primary focus on task-oriented grasping (TOG)—a critical capability that enables robots to determine not just *where* to grasp an object, but *how* to grasp it to accomplish a specific downstream task. His major contribution, the RTAGrasp framework, tackles a fundamental bottleneck in robotics: the scarcity of costly manual TOG annotations. By introducing a novel pipeline of Retrieval, Transfer, and Alignment, Liu’s work enables robots to learn precise grasping positions and directions directly from unstructured human demonstration videos, bypassing the need for expensive labeled datasets. This approach moves beyond prior methods that could only extract coarse grasping regions, achieving fine-grained, task-aware control. Though early in his career, his 2025 RTAGrasp paper has already garnered 5 citations, signaling strong interest from the manipulation community. Liu’s research bridges computer vision and robotics, offering a scalable path to teaching robots complex manipulation skills from human observation—a key step toward general-purpose service robots.
Research Focus
Key Achievements
Top Papers
- 1