Papers
10
Total Citations
499
H-Index
6
About
Yuheng Qiu is a robotics researcher whose work sits at the intersection of simultaneous localization and mapping (SLAM), robot perception, and the integration of deep learning with physics-based systems. He is perhaps best known for his contributions to the TartanAir dataset (2020), a landmark benchmark collected in photo-realistic simulated environments featuring dynamic objects, varying lighting, and adverse weather conditions — a resource that has accumulated over 390 citations and significantly shaped the trajectory of visual SLAM research. Building on this foundation, Qiu co-developed the SubT-MRS Dataset (2024, 46 citations), further pushing SLAM robustness toward all-weather, real-world scenarios. His work on PyPose (2023, 35 citations) demonstrates a broader ambition: bridging the generalization strengths of physics-based optimization with the representational power of deep learning for robot autonomy. More recently, his research has expanded into self-supervised neuro-symbolic learning through the Imperative Learning framework and open-set semantic scene understanding via RayFronts. Across his portfolio, Qiu consistently addresses a central challenge in robotics — building systems that generalize reliably beyond controlled, data-rich environments — making his work essential reading for researchers in autonomous navigation and robot learning.
Research Focus
Key Achievements
Top Papers
- 1TartanAir: A Dataset to Push the Limits of Visual SLAM365 citations · 2020
- 2SubT-MRS Dataset: Pushing SLAM Towards All-weather Environments46 citations · 2024
- 3PyPose: A Library for Robot Learning with Physics-based Optimization35 citations · 2023
- 4TartanAir: A Dataset to Push the Limits of Visual SLAM26 citations · 2020
- 5Unsupervised Online Learning for Robotic Interestingness With Visual Memory13 citations · 2021
- 6
- 7PyPose: A Library for Robot Learning with Physics-based Optimization3 citations · 2022
- 8
- 9
- 10