I'm an M.S. student in Intelligent Computing at National Cheng Kung University (AI System Lab). I work on visual localization and SLAM for robots, mostly with 3D Gaussian Splatting maps. I graduate in July 2027 and am open to full-time roles in robotics perception.
Current research
- Real-time 6-DoF visual localization against prebuilt 3DGS maps: sub-centimeter accuracy at 100+ FPS, about 47× faster than ULF-Loc (CVPR 2026 Highlight) with equal or better accuracy. Code release planned.
Projects
- FAST-LIVO-docker-env: a reproducible ROS Noetic environment for LiDAR-inertial-visual mapping (FAST-LIVO) on our lab's Ouster LiDAR + camera recordings.
- pros_car / ros2_yolo_integration: the autonomy layer for a ROS 2 mobile manipulator in a Unity digital twin, covering object retrieval, bridge crossing, and door opening.
- aoc-project: a team project building an INT8 DeiT-III accelerator. I worked on model quantization.
Tools Python · PyTorch · ROS 2 / ROS 1 · OpenCV · CUDA · Docker · Linux
Contact LinkedIn