Changxiang Liu

Papers

Total Citations

H-Index

About

Changxiang Liu is at the forefront of advancing visual-inertial navigation systems (VINS), a critical technology for robotics, autonomous driving, and computer vision. Their research focuses on enhancing the accuracy and robustness of monocular SLAM (Simultaneous Localization and Mapping) by integrating novel geometric features and deep learning techniques. Liu’s major contributions include the development of PE-VINS, which introduces point-edge features to improve performance over traditional point-line methods, achieving 7 citations since 2024. They also pioneered IMPS, an informative map point selection strategy that optimizes optimization-based VINS by minimizing reprojection errors, garnering 3 citations. Additionally, LPL-VIO demonstrates the fusion of deep learning-based point and line features into visual-inertial odometry, earning 2 citations. These works collectively address key challenges in real-world navigation, such as low-texture environments and dynamic scenes, pushing the boundaries of autonomous perception. Liu’s innovative approach—combining geometric rigor with learning-based adaptability—positions them as a rising leader in SLAM research, with their methods poised to influence next-generation autonomous systems.