CV
Basics
| Name | Tianyi Zhang |
| Status | PhD Student |
| tianyi.zhang@mrl.ethz.ch |
Education
-
2025.06 - now Zurich, Switzerland
-
2021.09 - 2025.03 Zurich, Switzerland
-
2016.09 - 2020.07 Harbin, China
Work
-
2020.09 - 2021.06 Shenzhen, China
Research Assistant
Southern University of Science and Technology (SUSTech)
Assisted in scientific research work related to robotics, including machine learning, computer vision, and path planning. Conducted research in image-based heuristic path planning algorithms for mobile robots. Participated in the preparation of ICRA 2021 conference and served as ICRA conference volunteer.
Publications
-
2024 Robust Incremental Structure-from-Motion with Hybrid Features
European Conference on Computer Vision ECCV 2024
Shaohui Liu*, Yidan Gao*, Tianyi Zhang*, Rémi Pautrat, Johannes L. Schönberger, Viktor Larsson, Marc Pollefeys
-
2021 Generative Adversarial Network based Heuristics for Sampling-based Path Planning
IEEE/CAA Journal of Automatica Sinica
Tianyi Zhang, Jiankun Wang, Max Q.-H. Meng
Skills
| Programming | |
| Python | |
| C++ | |
| ROS |
| Tools | |
| LATEX | |
| MarkDown | |
| Git |
| Language | |
| English (Fluent) | |
| Chinese (Native) |
Projects
-
2024.06 - 2025.02 Zurich, Switzerland
Gaussian Splatting with Geometric Constraints
Master Thesis, Supervised by Kunyi Li, Shaohui Liu, Tobias Fischer, Michael Niemeyer, Federico Tombari, Marc Pollefeys
- Developed a novel pose-free Gaussian Splatting approach within the Global Structure-from-Motion framework for sparse view, unposed, unordered images
-
2023.04 - 2023.11 Zurich, Switzerland
Exploiting Uncertainty for Building Enhanced 3D Maps
Semester Thesis, Supervised by Shaohui Liu, Marc Pollefeys
- Propagated the uncertainty for point and line mapping through analytical method to filter the degeneracy, identify unstable tracks, and help incorporate heterogeneous assumptions on the 2D observation noise. Developed an uncertainty-aware mechanism in hybrid point-line Structure-from-Motion and demonstrated the effectiveness