Mingfu Liang
Papers
1
Total Citations
9
H-Index
1
About
Mingfu Liang is a rising researcher at the intersection of computer vision, robotics, and embodied AI. His work centers on **active perception** and **open-world recognition**, where he tackles the fundamental challenge of enabling intelligent agents to explore and understand unfamiliar environments with both efficiency and caution. Liang is best known for his pioneering contributions to **evidential active recognition**, a framework that equips robots with the ability to not only identify known objects but also prudently handle novel or ambiguous observations. His 2024 paper, "Evidential Active Recognition: Intelligent and Prudent Open-World Embodied Perception," has already garnered 9 citations, reflecting its timely impact on the field. In this work, Liang demonstrates how agents can learn policies that intelligently seek out informative viewpoints while avoiding risky or uninformative actions—a crucial step toward deploying autonomous systems in real-world, unstructured settings. By blending evidential deep learning with active perception, he provides a principled way for robots to "know when they don't know," advancing the reliability of embodied AI. Liang’s research is shaping the next generation of perceptually aware robots, making him a notable voice in the push toward truly autonomous, open-world intelligence.
Research Focus
Key Achievements
Top Papers
- 1