Xuesong Chen
Papers
4
Total Citations
96
H-Index
3
About
Xuesong Chen is a researcher advancing the state of the art in 3D perception for autonomous systems. His primary research areas center on 3D object detection and multi-object tracking (MOT) using LiDAR point cloud sequences, with a direct impact on autonomous driving and service robotics. Chen’s most influential contribution is the MPPNet framework, a novel approach for multi-frame 3D temporal object detection that uses "proxy points" to intertwine features across time, achieving high accuracy and flexibility. This work has garnered 79 citations, establishing it as a key reference in the field. Building on this, Chen developed TrajectoryFormer, a transformer-based architecture for 3D MOT that moves beyond simple detection-box matching by predicting future trajectory hypotheses, a critical step for robust tracking in dynamic environments. Earlier in his career, he also contributed to mechanical engineering with a numerical simulation study on the structural integrity of heavy-duty brick palletizing robots. Through his focused work on temporal reasoning in point clouds, Chen is helping to build the reliable perception systems that will underpin next-generation autonomous vehicles and intelligent robots.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4