Leibing Xiao

Northwestern Polytechnical University, Shandong University

Papers

Total Citations

H-Index

About

Leibing Xiao is a rising researcher at the intersection of computer vision, robotics, and edge AI, whose work addresses critical real-world limitations in perception and manipulation systems. His primary research areas include multi-modal fusion, infrared imaging, and language-conditioned robotic grasping. Xiao’s most impactful contribution is **InfPose** (2023, 6 citations), a pioneering encoder-decoder CNN architecture that brings real-time multi-human pose estimation to infrared imagery on edge devices—solving the failure modes of RGB-based systems in nighttime and smoggy environments. This work is foundational for deploying robust perception in low-visibility conditions. Complementing this, his **Hierarchical Multi-Modal Fusion** framework (2024, 4 citations) tackles the complex challenge of language-conditioned grasping in cluttered scenes, enabling robots to interpret nuanced human instructions and generate precise grasping postures without relying on pre-trained object detectors. By fusing linguistic and visual cues hierarchically, Xiao’s approach advances intuitive human-robot interaction. Despite being early in his career, his publications demonstrate a clear trajectory toward practical, deployable intelligence—bridging the gap between high-level language understanding and low-level sensor processing. Xiao’s work is particularly notable for its focus on edge-computing efficiency, ensuring that sophisticated AI models can run on resource-constrained hardware, a key enabler for real-world robotics and surveillance applications.