Anqi Li | 李安齐

Hi, I am a junior undergraduate student at Yuanpei College, Peking University, majoring in Artificial Intelligence.
I'm currently visiting at Berkeley AI Research (BAIR), advised by Prof. Masayoshi Tomizuka. Previously, I was an intern at the Embodied AI startup Galbot, where I had the privilege to work with Prof. He Wang. I focus on embodied AI, particularly Vision-Language-Action (VLA) models and their applications in robotic navigation.
Anqi Li

News

  • 2026.01•🎉 UrbanVLA gets accepted to ICRA 2026.
  • 2026.01•🎉 NavFoM gets accepted to ICLR 2026.
  • 2025.11•I gave an invited talk at Yuanpei College, Peking University, introducing my work on urban micromobility.
  • 2025.08•🎉 TrackVLA gets accepted to CoRL 2025.

Research

SpanNav static
SpanNav animation

SPAN-Nav: Generalized Spatial Awareness for Versatile Vision-Language Navigation
Jiahang Liu*, Tianyu Xu*, Jiawei Chen*, Lu Yue*, Jiazhao Zhang*, Zhiyong Wang*, Minghan Li, Qisheng Zhao, Anqi Li, Qi Su, Zhizheng Zhang†, He Wang†
ArXiv Preprint
Paper Project page

SPAN-Nav is an end-to-end navigation model with universal 3D spatial awareness. It utilizes a single spatial token extracted from RGB video and a Spatial Chain-of-Thought mechanism to achieve robust path planning and obstacle avoidance in complex environments.

UrbanVLA static
UrbanVLA animation

UrbanVLA: A Vision-Language-Action Model for Urban Micromobility
Anqi Li*, Zhiyong Wang*, Jiazhao Zhang*, Minghan Li, Yunpeng Qi, Zhibo Chen, Zhizheng Zhang†, He Wang†
ICRA 2026
Paper Project page

UrbanVLA is a route-conditioned vision-language-action model for urban micromobility. It aligns high-level navigation routes with visual observations to enable scalable, long-horizon navigation.

NavFOM static
NavFOM animation

Embodied Navigation Foundation Model
Jiazhao Zhang*, Anqi Li*, Yunpeng Qi*, Minghan Li*, Jiahang Liu, Shaoan Wang, Haoran Liu, Gengze Zhou, Yuze Wu, Xingxing Li, Yuxin Fan, Wenjun Li, Zhibo Chen, Fei Gao, Qi Wu, Zhizheng Zhang†, He Wang†
ICLR 2026
Paper Project page

NavFoM is a cross-embodiment and cross-task navigation model trained on 8 million samples encompassing quadrupeds, drones, wheeled robots, and vehicles, spanning tasks including vision-and-language navigation, object searching, target tracking, and autonomous driving.

TrackVLA static
TrackVLA animation

TrackVLA: Embodied Visual Tracking in the Wild
Shaoan Wang*, Jiazhao Zhang*, Minghan Li, Jiahang Liu, Anqi Li, Kui Wu, Fangwei Zhong, Junzhi Yu, Zhizheng Zhang†, He Wang†
CoRL 2025
Paper Code Project page

TrackVLA is a vision-language-action model capable of simultaneous object recognition and visual tracking. It demonstrates robust tracking, long-horizon tracking, and cross-domain generalization across diverse challenging environments.

Experience

UC Berkeley

University of California, Berkeley

U.S. 2026.01 - Present

Visiting Student

Research Advisor: Prof. Masayoshi Tomizuka
Galbot

Galbot

China 2024.09 - 2026.01

Research Intern

Research Advisor: Prof. He Wang, Jiazhao Zhang

Peking University

Peking University

China 2023.09 - Present

Undergraduate Student

Major: Intelligence Science and Technology @ Yuanpei College
Minor: Innovation and Entrepreneurship Management @ Guanghua School of Management

Miscellaneous

Outside of research, I have several hobbies.

  • I enjoy playing basketball, and I am a huge fan of the NBA. My favorite player is Luka Dončić.
  • I love performing on the trombone. I am an active member of the Symphony Orchestra at my high school, Experimental High School attached to Beijing Normal University. Check out our recent performance.
  • I am a speedcuber. I participated in several speedcubing competitions held by the World Cube Association (WCA). Check out my WCA personal profile.

Webpage template adapted from Jon Barron.
Last updated: Jan., 2026