Papers
11
Total Citations
509
H-Index
7
About
Fanbo Xiang is a pioneering researcher at the intersection of robotics simulation, embodied AI, and robot learning, whose work has fundamentally advanced how intelligent systems are trained and evaluated in virtual environments. He is best known for creating SAPIEN, a physically realistic, part-based interactive simulation environment designed to bridge the gap between virtual training and real-world robotic deployment. With over 370 citations, SAPIEN stands as a landmark contribution to the field, providing researchers with articulated object datasets and simulation tools essential for developing home assistant robots. Building on this foundation, Xiang co-developed ManiSkill3, a GPU-parallelized robotics simulation and rendering framework that dramatically scales robot learning capabilities across diverse tasks and environments. His work on sensor simulation, particularly physics-grounded active stereo depth sensor modeling, addresses the critical sim-to-real transfer challenge by closing optical sensing domain gaps. Through the OCRTOC benchmark, he has also contributed cloud-based infrastructure for standardized evaluation of robotic grasping and manipulation. Across his research portfolio, Xiang consistently tackles the core challenge of making robot learning more generalizable, scalable, and practically transferable — work that continues to shape the trajectory of embodied AI research globally.
Research Focus
Key Achievements
Top Papers
- 1SAPIEN: A SimulAted Part-Based Interactive ENvironment373 citations · 2020
- 2
- 3SAPIEN: A SimulAted Part-based Interactive ENvironment22 citations · 2020
- 4
- 5
- 6O2O-Afford: Annotation-Free Large-Scale Object-Object Affordance Learning13 citations · 2021
- 7Part-Guided 3D RL for Sim2Real Articulated Object Manipulation9 citations · 2023
- 8
- 9
- 10