Xiangyang Xue
Papers
16
Total Citations
194
H-Index
6
About
Xiangyang Xue is a leading researcher at the intersection of computer vision, robotics, and human-robot interaction, with a primary focus on 6D object pose estimation, robotic grasping, and language-guided manipulation. His most impactful contribution is the SAR-Net (Shape Alignment and Recovery Network), which achieves category-level 6D object pose and size estimation from a single image without requiring real-world pose-annotated training data—a breakthrough that has garnered 86 citations. Xue has also pioneered work in language-guided robotic grasping, introducing methods like FLarG that enable robots to interpret fine-grained language instructions for precise object manipulation, and exploring freehand sketches as an alternative communication modality for grasp detection. His innovative approach extends to weakly-supervised liquid perception for robotic pouring (PourIt!), addressing a critical challenge in dynamic manipulation tasks. With over 180 citations across his top papers, Xue's research consistently pushes boundaries in making robots more perceptive and interactive. His recent work on amodal segmentation under occlusion (LAC-Net) and point cloud sampling with graph neural networks (GS-Net) demonstrates his ongoing commitment to solving fundamental perception challenges that enable more capable, human-aware robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4Language Guided Robotic Grasping with Fine-Grained Instructions16 citations · 2023
- 5
- 6
- 7
- 8
- 9
- 10GS-Net: Point cloud sampling with graph neural networks3 citations · 2025