Papers
6
Total Citations
137
H-Index
4
About
Guoyu Lu is a computer vision and robotics researcher whose work spans indoor localization, depth estimation, and embodied AI. He is best known for his pioneering contributions to image-based indoor localization, a challenging problem in environments where GPS and WiFi signals prove unreliable or insufficient. His 2015 paper on active transfer learning applied to thermal imaging-based indoor localization stands as his most influential work, amassing 77 citations and demonstrating innovative cross-modal approaches to positioning. Complementing this, his earlier multi-task and multi-view localization frameworks — including a 3D Structure-from-Motion (SfM) model-based system — established robust visual pipelines for determining location in complex indoor spaces, collectively contributing dozens of additional citations. More recently, Lu has extended his expertise into self-supervised depth estimation, exploring how camera models and vision-language frameworks can reduce costly ground-truth labeling requirements in monocular 3D perception. His 2024 and 2025 works reflect a forward-looking pivot toward embodied AI, integrating semantic language reasoning with visual depth understanding. Across his career, Lu has consistently addressed fundamental perception challenges in robotics, making his research valuable for students working at the intersection of computer vision, mobile computing, and autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2Indoor localization via multi-view images and videos21 citations · 2017
- 3
- 4Image-based indoor localization system based on 3D SfM model14 citations · 2014
- 5Embodiment: Self-Supervised Depth Estimation Based on Camera Models4 citations · 2024
- 6Vision-Language Embodiment for Monocular Depth Estimation3 citations · 2025