Xibin Song

Baidu (China), National Engineering Laboratory of Deep Learning Technology and Application

Papers

Total Citations

104

H-Index

About

Xibin Song is a leading researcher in robotics and computer vision, whose work bridges the gap between autonomous manipulation and 3D scene understanding. His most impactful contributions center on developing intelligent systems for construction machinery and dense environmental perception. In his highly cited 2021 work, "Optimization-Based Framework for Excavation Trajectory Generation" (58 citations), Song pioneered a novel approach that expands the optimization space for autonomous excavators, moving beyond traditional oversimplified trajectory parameterizations to enable more efficient and task-specific digging operations. This foundational work has significant implications for automating heavy machinery in construction and mining. Song has also made substantial advances in depth perception. His 2023 paper, "MFF-Net: Towards Efficient Monocular Depth Completion With Multi-Modal Feature Fusion" (39 citations), tackles the critical challenge of generating dense depth maps from sparse sensor data, proposing a novel multi-modal fusion architecture that overcomes limitations in feature extraction. Further extending this line of research, his 2022 work "DiT-SLAM" introduces a real-time dense visual-inertial SLAM system that leverages implicit depth representations from deep neural networks, enabling robots to build rich, continuous maps of their environments during navigation. Through these innovations, Song is shaping the future of autonomous systems that can both perceive and act in complex, unstructured environments.