Xiaodan Liang

Sun Yat-sen University, Carnegie Mellon University

Papers

14

Total Citations

611

H-Index

9

About

Xiaodan Liang is a prominent researcher whose work spans computer vision, embodied artificial intelligence, and human-robot interaction. Beginning with foundational contributions in RGB-D scene understanding — most notably the LSTM-CF framework (2016, 188 citations), which unified context modeling and fusion for depth-aware scene labeling — Liang progressively expanded into the ambitious frontier of embodied AI, where intelligent agents must perceive, reason, and act within complex three-dimensional environments. A defining thread throughout Liang's career is enabling machines to understand and anticipate human behavior. Her 2018 work on teaching robots to predict human motion (125 citations) demonstrated how deep learning could equip robots with proactive collaborative capabilities. This evolved into landmark contributions in vision-and-language navigation, including the SOON benchmark (115 citations) and NavCoT, which leverages large language models for disentangled navigational reasoning. Her 2025 comprehensive survey on embodied AI (77 citations) signals her role as a field synthesizer, bridging cyberspace and physical-world intelligence. Liang's body of work reflects a coherent vision: building robots that see, communicate, predict, and navigate with human-like fluency — making her an essential voice in the ongoing pursuit of artificial general intelligence.

Research Focus

Key Achievements

9
H-Index
14
Papers
611
Total Citations
44
Avg Citations/Paper
🏆 Most Cited Paper
LSTM-CF: Unifying Context Modeling and Fusion with LSTMs for RGB-D Scene Labeling
188 citations · 2016
📈 Most Prolific Year: 2021 (5 Papers)
🤝 Key Collaborators: 61
🏛 Institutions: Sun Yat-sen University, Carnegie Mellon University

Top Papers

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 14 days ago