Junshan Huang

Junshan Huang

PhD Student

Rutgers University

Research Interests

Embodied AI
Kitchen Automation
Dexterous Manipulation

About

I am a PhD student at the School of Computer Science, Rutgers University, advised by Prof. Jingjin Yu.

Prior to this, I obtained a BSc degree with Honor Degree (top 5%) in Artificial Intelligence from the University of Science and Technology of China (USTC).

My research focuses on developing more capable and trustworthy robot brains that allow robots to work efficiently and reliably in household environments.

The kitchen, as a compact yet highly representative domestic space, serves as an ideal testbed for this research.

I organize kitchen-related tasks into three main categories:

  • Ingredient Preparation: Preparing raw ingredients for cooking.
  • Cooking: Autonomously turning prepared ingredients into delicious meals.
  • Post-Meal Tidying: Restoring the kitchen to a clean state for the next round of cooking.

We have shipped OrganizeIt as part of our work on post-meal tidying, and are also making progress on the other two categories.

Selected Publications

View All →
OrganizeIt: Compiling Generated Images into Tidy Tabletop Arrangements

OrganizeIt: Compiling Generated Images into Tidy Tabletop Arrangements

Junshan Huang, Xiyu Ke, Jintong Li, Litao Liu, Duo Zhang, Jingjin Yu

Under Review

Achieve zero-shot, CAD-free, general tabletop organization with future image guidance and relational structure guarantees.

Paper
High-Performance Dual-Arm Task and Motion Planning for Tabletop Rearrangement

High-Performance Dual-Arm Task and Motion Planning for Tabletop Rearrangement

Duo Zhang, Junshan Huang, Jingjin Yu

Proceedings of the IEEE International Conference on Robotics and Automation (ICRA)

We propose SDAR, a dual-arm rearrangement planner that efficiently solves complex, entangled object manipulation tasks through synchronous task and motion planning.

PaperProjectCode
VLA-OS: Structuring and Dissecting Planning Representations and Paradigms in Vision-Language-Action Models

VLA-OS: Structuring and Dissecting Planning Representations and Paradigms in Vision-Language-Action Models

Chongkai Gao, Zixuan Liu, Zhenghao Chi, Junshan Huang, Xin Fei, Yiwen Hou, Yuxuan Zhang, Yudi Lin, Zhirui Fang, Lin Shao

Advances in Neural Information Processing Systems (NeurIPS)

Unified benchmarking reveals visually grounded hierarchical planning excels in VLAs.

PaperProjectCode

News

2026-02

🎉 Our paper SDAR was accepted by ICRA 2026!

2025-09

🎉 Our paper VLS-OS accepted by NeurIPS 2025!

2025-09

Starting my PhD at the Rutgers University~

2025-06

🎓 Graduated from USTC with a honor degree! (Top 5%)

2025-06

Awarded the AI Talent Program Scholarship from USTC.

2023-10

Awarded the Shenzhen Stock Exchange Scholarship.

2023-10

We win the 1st place in the ACM REACT 2023 Multimodal Challenge!