PERSONAL LAB NOTES / SHANGHAI

Building clearer machines for the road.

I am Zedong Yu (Yu Zetong), a researcher and engineer exploring how perception, world models, and multimodal agents can work together for autonomous driving.

Explore the knowledge base Read the latest notes -> View publications ->

Currently exploring
AD perception -> world models -> LLM agents

PUBLIC RESEARCH SNAPSHOT

Verified from GitHub and Google Scholar / Aug 2026

65Citations 3h-index 1i10-index 20Public repos

Two ways in

01 / KNOWLEDGE BASE

Make ideas reusable.

Evergreen notes, concept maps, and practical references for radar perception and multimodal intelligence.

Browse topics -> 02 / BLOG

Follow the working trail.

Short research notes, engineering logs, and the questions that shape the next experiment.

Read the journal ->

Selected work

01

Radar perception map

A curated index of deep-learning-based radar perception papers, code, and directions.

87 stars / 9 forks
02

CRFNet on TensorFlow 2

A Windows and TensorFlow 2.4.1 reproduction of the CRFNet official implementation.

43 stars / 5 forks
03

ARS 408 ROS Toolkit

Practical tooling for working with radar data in a ROS-based autonomous-driving stack.

24 stars / 10 forks
04

SAF-FCOS reproduction

An open reproduction of the official SAF-FCOS repository for 3D perception research.

13 stars / 3 forks

A SMALL PROMISE

Publish what is useful. Keep what is unfinished. Link the two.

Zedong Yu (于泽东)

  • πŸ“ Shanghai
  • πŸ’Ό AD perception β†’ AD World model & LLM agents & VLM
  • πŸ”— GitHub

Focus

Research and engineering interests:

  • AD World Model
  • LLM Agents
  • Vision Language Models (VLM)

Recent Posts

See the blog for learning notes and research thoughts.