Shiliang Pu
Papers
4
Total Citations
70
H-Index
3
About
Shiliang Pu is a leading researcher in robotic vision, with a primary focus on domain adaptation, lifelong learning, and sensor calibration. His most impactful work, "Self-Supervised Noisy Label Learning for Source-Free Unsupervised Domain Adaptation" (2022, 51 citations), addresses a critical challenge in robot vision: enabling neural networks pre-trained on source domains to adapt to target domains without requiring source data—a practical necessity given storage constraints. This contribution has significant implications for deploying robots in dynamic, real-world environments. Pu also played a key role in the IROS 2019 Lifelong Robotic Vision Challenge, co-authoring both the competition overview (12 citations) and the final report (2 citations), which introduced the OpenLORIS-object dataset and benchmarked top methods from over 150 teams. His work on "Simultaneous Intrinsic and Extrinsic Calibration of a Visual-Odometric Sensor System" (2020, 5 citations) further demonstrates his expertise in fusing multi-sensor data for precise robot navigation. Through these contributions, Pu is advancing the frontier of autonomous, continuously learning robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4