Shuting Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Shuting Zhang has a total of 8 co-authors. Here are the top co-authors by number of shared publications:
- Bocheng Zhu
School of Electronics, Peking University, Beijing 100871, China
2 shared publications - Guangchen Li
Peking University
2 shared publications - Haowen Wang
Peking University
2 shared publications - Fangzhou Tang
School of Electronics, Peking University, Beijing, China
1 shared publication - Junren Sun
School of Artificial Intelligence, China University of Mining and Technology-Beijing, Beijing 100091, China
1 shared publication - Xin Meng
Peking University
1 shared publication
Latest Publications
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (10) , 11936-11943
AeroDiffusion: Real-Time Quadrotor Trajectory Planning in Dynamic Environments via Conditional Diffusion
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (10) , 11936-11943
AeroDiffusion: Real-Time Quadrotor Trajectory Planning in Dynamic Environments via Conditional Diffusion
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (10) , 23015-23031
MC-GVIO: Motion Consistency Aided GNSS-Visual-Inertial Odometry for Land Vehicle Navigation in Urban Environments
- 1 October 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 75 (10) , 23015-23031
MC-GVIO: Motion Consistency Aided GNSS-Visual-Inertial Odometry for Land Vehicle Navigation in Urban Environments
- 2 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 74 (10) , 15441-15455
Enhanced LiDAR Odometry for Autonomous Vehicular Positioning System Using Local Feature Enhancement and Global Motion Constraint
LiDAR Odometry (LO) has played a critical role in vehicular positioning technology for autonomous vehicle systems and undergone significant advancements. However, endto-end deep learning-based LO techniques applied in vehicular positioning fall short of traditional approaches in performance. A novel end-to-end LO method named LAGLO is proposed, which is in accordance with the analysis of the range view of LiDAR point clouds. LAGLO comprises two key components: an Enhanced Temporal Network (ETN) with Local Feature Enhancement (LFE) and Global Motion Constraint (GMC), integrated with Motion Mode Attention (MMA). The ETN addresses the challenge of capturing temporal dependencies, thereby enabling more accurate predictions of motion. The GMC improves global vehicle localization by introducing motion-based constraints, which reduces drift in pose estimation. The MMA improves feature focus by capturing motion modes, thereby improving localization performance. LAGLO is the first method to combine traditional estimation techniques (e.g., optical flow and background segmentation) with deep learning LO using point cloud range views. In addition, LAGLO is the first deep learningbased LO method to introduce motion constraint. Furthermore, to address the limitations for applications in vehicular positioning systems, LAGLO supports both supervised and self-supervised training modes. For the self-supervised mode, a series of optimization strategies have been developed to enhance the algorithm's robustness in vehicle positioning. Experiments involving the public KITTI dataset validate the superior performance of LAGLO. LAGLO represents the latest advancement in end-toend deep learning-based LO techniques for vehicle positioning using range views, and its training paradigm provides new insights for developing higher accuracy and reliability in automotivegrade deep learning-based LO for autonomous vehicular systems.
- 2 May 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 74 (10) , 15441-15455
Enhanced LiDAR Odometry for Autonomous Vehicular Positioning System Using Local Feature Enhancement and Global Motion Constraint
LiDAR Odometry (LO) has played a critical role in vehicular positioning technology for autonomous vehicle systems and undergone significant advancements. However, endto-end deep learning-based LO techniques applied in vehicular positioning fall short of traditional approaches in performance. A novel end-to-end LO method named LAGLO is proposed, which is in accordance with the analysis of the range view of LiDAR point clouds. LAGLO comprises two key components: an Enhanced Temporal Network (ETN) with Local Feature Enhancement (LFE) and Global Motion Constraint (GMC), integrated with Motion Mode Attention (MMA). The ETN addresses the challenge of capturing temporal dependencies, thereby enabling more accurate predictions of motion. The GMC improves global vehicle localization by introducing motion-based constraints, which reduces drift in pose estimation. The MMA improves feature focus by capturing motion modes, thereby improving localization performance. LAGLO is the first method to combine traditional estimation techniques (e.g., optical flow and background segmentation) with deep learning LO using point cloud range views. In addition, LAGLO is the first deep learningbased LO method to introduce motion constraint. Furthermore, to address the limitations for applications in vehicular positioning systems, LAGLO supports both supervised and self-supervised training modes. For the self-supervised mode, a series of optimization strategies have been developed to enhance the algorithm's robustness in vehicle positioning. Experiments involving the public KITTI dataset validate the superior performance of LAGLO. LAGLO represents the latest advancement in end-toend deep learning-based LO techniques for vehicle positioning using range views, and its training paradigm provides new insights for developing higher accuracy and reliability in automotivegrade deep learning-based LO for autonomous vehicular systems.