Seth Z. Zhao

profile_pic_2022.png

I am a third-year Computer Science PhD student at UCLA, advised by Professor Bolei Zhou and Professor Jiaqi Ma. My research is supported by Qualcomm Innovation Fellowship.

Previously, I received my M.S. and B.A. in Computer Science at UC Berkeley where I was fortunate to work with Professor Masayoshi Tomizuka, Professor Allen Yang, and Professor Constance Chang-Hasnain. I was also a research intern at Qualcomm and Honda Research Institute (HRI) before.

My current research interest lies at the building efficient and scalable compound agentic AI systems for autonomous driving and embodied AI.

News

May 14, 2026 Honored to receive Qualcomm Innovation Fellowship 2026 together with my fantastic teammate Zewei Zhou!
May 01, 2026 Paper on AgentAlign: Misalignment-Aware Multi-Agent Perception for Resilient Cross-Agent Feature Alignment has been accepted by T-ITS 2026.
Nov 03, 2025 Paper on Driving with Regulation: Trustworthy and Interpretable Decision-Making for Autonomous Driving with Retrieval-Augmented Reasoning has been accepted by AAAI 2026 (AISI Track).
Oct 01, 2025 Check out our research activities at ICCV 2025 in Hawaii: (1) present two main conference papers, V2XPnP and TurboTrain; (2) co-organize Tutorial on Beyond Self-Driving: Exploring Three Levels of Driving Automation; (3) workshop papers presentations of (i) AutoVLA at ICCV 2nd WDFM-AD Workshop, (ii) V2XPnP at ICCV 2025 DriveX Workshop as oral presentation, (iii) TurboTrain and QuantV2X at ICCV 2025 X-Sense Workshop as oral presentations.
Sep 18, 2025 Paper on AutoVLA: A Vision-Language-Action Model for End-to-End Autonomous Driving with Adaptive Reasoning and Reinforcement Fine-Tuning has been accepted by NeurIPS 2025.
Aug 15, 2025 Invited talk on Towards Efficient and Scalable Multi-Agent Autonomous Driving Systems at Qualcomm PhD Lecture Series.
Jun 26, 2025 Two papers have been accepted by ICCV 2025! Check out V2XPnP: Vehicle-to-Everything Spatio-Temporal Fusion for Multi-Agent Perception and Prediction and TurboTrain: Towards Efficient and Balanced Multi-Task Learning for Multi-Agent Perception and Prediction.
Jun 16, 2025 Paper on CooPre: Cooperative Pretraining for V2X Cooperative Perception has been accepted by IROS 2025 as oral presentation. This paper has been awarded with Best Paper Award at the CVPR 2025 DriveX Workshop and presented as oral presentation.
May 01, 2025 Paper on WOMD-Reasoning: A Large-Scale Dataset for Interaction Reasoning in Driving has been accepted by ICML 2025.
Feb 10, 2025 Invited talk on UCLA Smart Intersection for Vehicle-to-Everything Application at 1st UCLA-UCB Joint Workshop: Towards Next Generation of Open-Source Simulator for Embodied AI and Autonomous Driving.
Feb 01, 2025 Paper on Towards Autonomous Micromobility through Scalable Urban Simulation has been accepted by CVPR 2025 (Highlight).
Jan 01, 2025 Our team of UCLA Mobility Lab secured the Tier 1 in the U.S.DOT Intersection Safety Challenge with a $750k award!
Jul 01, 2024 Paper on Pre-training on Synthetic Driving Data for Trajectory Prediction has been accepted by IROS 2024.

Selected Publications

  1. Preprint
    mdrive.png
    MDrive: Benchmarking Closed-Loop Cooperative Driving for End-to-End Multi-agent Systems
    Marco Coscoy*, Zewei Zhou*, Seth Z. Zhao*, Henry Wei, Angela Magtoto, Johnson Liu, Rui Song, Walter Zimmer, Zhiyu Huang, Chen Tang, Bolei Zhou, and Jiaqi Ma
    2026
    HTML Code
  2. Preprint
    bridgesim.png
    BridgeSim: Unveiling the OL-CL Gap in End-to-End Autonomous Driving
    Seth Z. Zhao*, Luobin Wang*, Hongwei Ruan, Yuxin Bao, Yilan Chen, Ziyang Leng, Abhijit Ravichandran, Honglin He, Zewei Zhou, Xu Han, Abhishek Peri, Zhiyu Huang, Pranav Desai, Henrik Christensen, Jiaqi Ma, and Bolei Zhou
    2026
    HTML Code
  3. Preprint
    quantv2x.png
    QuantV2X: A Fully Quantized Multi-Agent System for Cooperative Perception
    Seth Z. Zhao*, Huizhi Zhang*, Zhaowei Li, Juntong Peng, Anthony Chui, Zewei Zhou, Zonglin Meng, Hao Xiang, Zhiyu Huang, Fujia Wang, Ran Tian, Chenfeng Xu, Bolei Zhou, and Jiaqi Ma
    2025
    HTML Code
  4. NeurIPS 2025
    autovla.png
    AutoVLA: A Vision-Language-Action Model for End-to-End Autonomous Driving with Adaptive Reasoning and Reinforcement Fine-Tuning
    Zewei Zhou, Tianhui Cai, Seth Z. Zhao, Yun Zhang, Zhiyu Huang, Bolei Zhou, and Jiaqi Ma
    In NeurIPS, 2025
    HTML Code
  5. ICCV 2025
    turbotrain.png
    TurboTrain: Towards Efficient and Balanced Multi-Task Learning for Multi-Agent Perception and Prediction
    Zewei Zhou*, Seth Z. Zhao*, Tianhui Cai, Zhiyu Huang, Bolei Zhou, and Jiaqi Ma
    In ICCV, 2025
    HTML Code
  6. IROS 2025 Oral
    coopre.png
    CooPre: Cooperative Pretraining for V2X Cooperative Perception
    Seth Z. Zhao, Hao Xiang, Chenfeng Xu, Xin Xia, Bolei Zhou, and Jiaqi Ma
    In IROS Oral, 2025
    HTML Code
  7. IROS 2024
    iros2024.png
    Pre-training on Synthetic Driving Data for Trajectory Prediction
    Yiheng Li*, Seth Z. Zhao*, Chenfeng Xu, Chen Tang, Chenran Li, Mingyu Ding, Masayoshi Tomizuka, and Wei Zhan
    In IROS, 2024
    HTML Code