Scholar
Letian Wang
Google Scholar ID: HEzCWisAAAAJ
University of Toronto | Carnegie Mellon University | UC Berkeley
Citations & Impact
Citations
1,378
H-index
12
i10-index
13
Publications
20
Co-authors
19
list available
Publications
Resume (English only)
Academic Achievements
- Awardee of Qualcomm Fellowship 2025
- Winner of the 2022 CARLA Autonomous Driving Challenge
- Best Paper Award Honorable Mention at IEEE RA-L 2021
- First Prize in the National Challenge Cup 2017 (China’s premier university-level science and technology competition)
- Authored the book 'Social Interactions for Autonomous Driving', published by Foundations and Trends in Robotics
- Multiple papers accepted at top-tier conferences: ICCV (1), IROS (2), ACM-MM (1), NeurIPS (DistillNeRF, Visual CoT Spotlight), CVPR (LmDrive, SmartRefine, ReasonNet), ICLR (SmartPretrain), RSS (ASAP-RL)
- LmDrive is the first work to integrate LLMs into closed-loop end-to-end autonomous driving
- Interfuser ranked first on the CARLA autonomous driving leaderboard
- Recipient of the Ontario Graduate Scholarship, Canada
- Co-founded a startup in industrial UAVs
Research Experience
- 2023–2024: Research Intern at NVIDIA Research Autonomous Vehicle Research Group, collaborating with Peter Karkus, Seung Wook Kim, Boris Ivanovic, Yue Wang, Sanja Fidler, and Marco Pavone
- 2020–2022: Research at CMU Robotics Institute with Prof. Changliu Liu and Yeping Hu
- 2019–2020: Research at UC Berkeley with Prof. Masayoshi Tomizuka, Liting Sun, and Wei Zhan
- 2018: Research at HKUST Robotics Institute with Prof. Fu Zhang
- Starting 2025: Joining Google DeepMind
Background
- Ph.D. student at the University of Toronto, affiliated with the Toronto Robotics and AI Lab
- Member of the Vector Institute, founded by Prof. Geoffrey Hinton
- Research interests lie at the intersection of robotics, machine learning, and computer vision
- Special focus on 3D/4D vision, multimodal learning, LLM agents, end-to-end self-driving, human-robot interaction, and motion forecasting
- Recently focused on developing generalizable decision-making and scalable perception systems powered by foundation models and data-scalable learning paradigms, with safety as the top priority
