Haitao Song

Xi'an High Tech University, Chinese Academy of Sciences, China Coal Technology and Engineering Group Corp (China)

Papers

Total Citations

H-Index

About

Haitao Song’s research bridges the critical intersection of computer vision and robotics, with a focus on enhancing perception and autonomy in challenging real-world environments. His work spans robust visual odometry, surgical tool analysis, and industrial automation. Song’s most cited paper, “Robust RGB-D visual odometry based on edges and points” (2018, 22 citations), introduces a method that fuses edge and point features to improve camera tracking accuracy in low-texture or poorly lit scenes—a fundamental challenge in robotics. In the surgical domain, his paper “Multiscale matters for part segmentation of instruments in robotic surgery” (2020, 6 citations) tackles the difficult task of distinguishing instrument parts with similar textures, proposing an end-to-end recurrent model that leverages multiscale features for finer segmentation. More recently, Song has applied vision-guided strategies to industrial robotics, as seen in “A novel vision-guided strategy for accurately delivering the drill pipe of the horizontal directional drilling rigs” (2023, 3 citations), where a 2D vision sensor enables precise pipe delivery, reducing labor intensity in coal mining. Collectively, Song’s contributions demonstrate a commitment to making robots more perceptive and reliable—from the operating room to the drilling site.