Xiaolin Fang
Papers
5
Total Citations
56
H-Index
3
About
Xiaolin Fang is a robotics researcher whose work sits at the intersection of task and motion planning (TAMP), perception under uncertainty, and imitation learning. Her major contributions focus on enabling robots to manipulate unknown objects in long-horizon, partially observable environments. In her highly cited 2022 paper (35 citations), she introduced a general-purpose TAMP system that combines engineered and learned modules to estimate object affordances from RGB images, allowing robots to plan and execute complex manipulation tasks without prior object models. Fang also pioneered the use of diffusion models as samplers for TAMP under partial observability (DiMSam, 2024, 10 citations), bridging the gap between generative models’ distribution-capturing strengths and TAMP’s constraint-reasoning capabilities. Her work on keypoint abstraction (KALM, 2025, 6 citations) leverages large models to enable object-relative imitation learning that generalizes across novel configurations and tasks. Additionally, she developed uncertainty-aware object segmentation (UncOS, 2024) for embodied interactive perception. With a growing citation impact and a clear trajectory toward more general, uncertainty-robust robot autonomy, Fang is establishing herself as a leading voice in next-generation manipulation systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4Embodied Uncertainty-Aware Object Segmentation3 citations · 2024
- 5