Event3R: Asynchronous-to-Global 3D Reconstruction from Event Camera via Spatial-Temporal Feature Aggregation
A feed-forward event-camera reconstruction framework that predicts globally aligned 3D point clouds from asynchronous event streams.
I am an incoming Ph.D. student at Westlake University, where I will join the Spatial Intelligence and Robotics Lab under the supervision of Prof. Peidong Liu.
My previous research focused on event-based 3D perception, reconstruction, and SLAM. During my Ph.D. study, I am shifting toward Embodied AI, with a current focus on local manipulation for humanoid robots. I am broadly interested in building robotic systems that can perceive geometry, reason about nearby objects, and act reliably in physical environments.
* denotes equal contribution. † denotes corresponding author.
A feed-forward event-camera reconstruction framework that predicts globally aligned 3D point clouds from asynchronous event streams.
An event-depth Gaussian Splatting SLAM system with robust patch-based tracking for long event streams and high-fidelity reconstruction.
Email: shenhaotian@westlake.edu.cn
GitHub: Tesla-SHT
I also use this site to keep academic notes, technical write-ups, and occasional personal records. It is a small place for thinking aloud while I learn.