Tarot: Targeted data selection via optimal transport
Lan Feng
•
•
1 min readProfessional Summary
I am currently a Ph.D. candidate advised by Prof. Alexandre Alahi at EPFL, Switzerland. Before that, I received my master’s degree in Robotics, Systems and Control from ETH Zurich and my bachelor’s degree from Wuhan University.
During my master studies, I worked on human-robot handover at the VIT Lab at ETH Zurich with Prof. Otmar Hilliges. Prior to that, I was engaged in applying generative models to traffic simulation with Prof. Bolei Zhou from UCLA.
My research interests focus on applying data-centric techniques to LLMs, robotics, and computer vision to improve performance with less computational resources.
Education
PhD Robotics
EPFL, Switzerland
MS Robotics
ETH Zurich, Switzerland
BE Navigation Engineering
Wuhan University, China
Interests
Data Selection & Curation
Autonomous Driving
Video Generation & Understanding
Recent News 📰
- Oct 7, 2025 — 🚀 Our work RAP: 3D Rasterization Augmented Planning is released!
- June 1, 2025 - 🏆 I won the 1st place in the 2025 Waymo End-to-End Planning Challenge! Thanks to Alex & Waymo.
- Dec 3, 2024 — More data is not all you need? TAROT is now on arXiv! See how we use less data for better performance.
- Jul 17, 2024 — UniTraj is accepted to ECCV 2024 🔥! Thanks to all my collaborators.
- Mar 25, 2024 — Our work on scalable vehicle trajectory prediction: UniTraj is now on Arxiv. Code is released! 🚀
- Jan 29, 2024 — My thesis on SynH2R has been accepted to ICRA 2024!
- Nov 1, 2023 — I came to the VITA lab at EPFL for an internship. I will work with Prof. Alexandre Alahi on autonomous driving.
Featured Publications
Read more
Synh2r: Synthesizing hand-object motions for learning human-to-robot handovers
Lan Feng
•
•
1 min read
Unitraj: A unified framework for scalable vehicle trajectory prediction
Lan Feng
•
•
1 min read
Trafficgen: Learning to generate diverse and realistic traffic scenarios
Lan Feng
•
•
1 min readRecent Publications
(2025).
OmniTraj: Pre-Training on Heterogeneous Data for Adaptive and Zero-Shot Human Trajectory Prediction.
arXiv preprint arXiv:2507.23657.
(2025).
Tarot: Targeted data selection via optimal transport.
In ICML 2025.
(2025).
Weak-for-Strong: Training Weak Meta-Agent to Harness Strong Executors.
arXiv preprint arXiv:2504.04785.
(2024).
Synh2r: Synthesizing hand-object motions for learning human-to-robot handovers.
In ICRA 2024.
(2024).
Unitraj: A unified framework for scalable vehicle trajectory prediction.
European Conference on Computer Vision.