Zhuohao Yan
First-year Ph.D. student at Fudan University. I work on humanoid motion control and vision-language-action (VLA) models.
About
I am a first-year Ph.D. student at Fudan University, advised by Prof. Wenchao Ding at MagicLab. I received my M.S. in Navigation, Guidance, and Control (GPA 93/100) and my B.S. in Navigation Engineering (GPA 3.9/4.0, top 3%) from the School of Geodesy and Geomatics at Wuhan University, where I was advised by Prof. Xingxing Li.
My current research is on humanoid motion control and vision-language-action models. During my master’s I worked on tightly coupled multi-object tracking and LiDAR-inertial odometry, and on monocular 3D tracking with state-space models.
Selected work
S3MOT: Monocular 3D Object Tracking with Selective State Space Model
arXiv:2504.18068
HSSM, VeloSSM, and FCOE fuse appearance, motion, and spatial cues into a differentiable association model. 76.86 HOTA at 31 FPS on the KITTI test set.
LIO-LOT: Tightly-Coupled Multi-Object Tracking and LiDAR-Inertial Odometry
IEEE Transactions on Intelligent Transportation Systems, 2024
Puts dynamic objects and LIO in one factor graph, with hierarchical coupling by tracking confidence. Evaluated on KITTI and nuScenes.