Yurun Chen

Shot by Jimmie Park, 2022

Human-Centered Embodied Intelligence · Robot Self-Understanding & Reliable Action

I am part of the joint doctoral program of Shanghai Jiao Tong University and Eastern Institute of Technology, Ningbo, advised by Prof. Wenjun Zeng and Prof. Wentao Zhu.

Before starting my PhD, I received my master’s degree in Data Science and Information Technology from Tsinghua University, where I was advised by Prof. Ruqi Huang. Earlier, I received my bachelor’s degree in Software Engineering from Dalian University of Technology.

My goal is to build human-centered embodied agents that understand their own bodies, act reliably in the physical world, and collaborate naturally with people. My current work develops three connected foundations toward this goal:

Humanoid self-representation: learning self–other distinction and predictive 3D self-models from proprioceptive–visual experience.
Reliable embodied-agent systems: integrating language models, robot policies, physical tools, structured world state, and action verification for long-horizon tasks and failure recovery.
Humanoid motion and manipulation: deploying locomotion, grasping, and pick-and-place capabilities on the Unitree G1 humanoid robot.

My broader interests include AI for remote sensing, interpretable scientific machine learning, and generative 3D rigging.

See my research projects, publications, or CV for details.

News

Aug 03, 2026	We released the arXiv preprint Towards the Harness of Embodied Agents and the Thea project page.
Jun 11, 2026	Our humanoid self–other distinction work is available on arXiv. Project
May 15, 2026	Our physics-guided KAN paper on interpretable soil-moisture retrieval was published in Remote Sensing of Environment.

Selected Publications

arXiv
Proprioceptive-visual Correspondence Enables Self-other Distinction in Humanoid Robots

Yurun Chen, Tianyuan Gao, Yizhong Ge, Shikun Ban, Yizhou Wang, Hongkai Xiong, Wenjun Zeng, and Wentao Zhu

Jun 2026

Abs arXiv Bib HTML PDF Website

A humanoid robot learns self-other distinction from proprioceptive-visual correspondence without identity labels or a predefined kinematic model. The resulting distinction supervises a predictive 3D self-model and supports target reaching, collision-aware motion planning, and human-to-robot motion retargeting.
@article{chen2026selfother, title = {Proprioceptive-visual Correspondence Enables Self-other Distinction in Humanoid Robots}, author = {Chen, Yurun and Gao, Tianyuan and Ge, Yizhong and Ban, Shikun and Wang, Yizhou and Xiong, Hongkai and Zeng, Wenjun and Zhu, Wentao}, year = {2026}, month = jun, }
RSE
A Novel Soil Moisture Retrieval Method via Combining Radiative Transfer Model and Machine Learning

Yurun Chen, Cheng Tong, Josh Qixuan Sun, Yulin Shangguan, Xiaodong Deng, Mark Crowley, Hongquan Wang, Yang Ye, Haijun Bao, and Ruqi Huang

Remote Sensing of Environment. Published. JIF 12.3 (2025), JCR Q1 (SCIE). Physics-guided, interpretable scientific machine learning with KAN , May 2026

Abs DOI Bib HTML PDF

This work combines a radiative transfer model with a Kolmogorov-Arnold Network to learn explicit soil-moisture retrieval formulas. The framework preserves physical structure, exposes relationships among key variables, and improves interpretability and real-data generalization over a conventional multilayer perceptron.
@article{chen2026soil, title = {A Novel Soil Moisture Retrieval Method via Combining Radiative Transfer Model and Machine Learning}, author = {Chen, Yurun and Tong, Cheng and Sun, Josh Qixuan and Shangguan, Yulin and Deng, Xiaodong and Crowley, Mark and Wang, Hongquan and Ye, Yang and Bao, Haijun and Huang, Ruqi}, journal = {Remote Sensing of Environment}, volume = {338}, pages = {115378}, year = {2026}, month = may, doi = {10.1016/j.rse.2026.115378}, }
arXiv
Towards the Harness of Embodied Agents

Qi Wang, Tianyi Wang, Chengyang Li, Shikun Ban, Yurun Chen, Yizhong Ge, Jason Qin, Chengtai Li, and Wentao Zhu

arXiv preprint · Technical Report. My contribution: Unitree G1 locomotion and manipulation deployment, closed-loop tool use, persistent scene state, and action-outcome evaluation , Aug 2026

Abs arXiv Bib HTML PDF Website

Thea is an embodied-agent harness that unifies language models, robot policies, and physical tools in a closed-loop tool-calling framework. Persistent scene graphs and independent evaluators provide structured world state, action termination, outcome verification, and failure diagnosis for long-horizon tasks.
@techreport{wang2026thea, title = {Towards the Harness of Embodied Agents}, author = {Wang, Qi and Wang, Tianyi and Li, Chengyang and Ban, Shikun and Chen, Yurun and Ge, Yizhong and Qin, Jason and Li, Chengtai and Zhu, Wentao}, institution = {Technical Report}, year = {2026}, month = aug, }
ICCV
ARMO: Autoregressive Rigging for Multi-Category Objects

Mingze Sun, Shiwei Mao, Keyi Chen, Yurun Chen, Shunlin Lu, Jingbo Wang, Junting Dong, and Ruqi Huang

In Proceedings of the IEEE/CVF International Conference on Computer Vision, Oct 2025

Abs Bib HTML PDF Website

ARMO predicts 3D joint locations and skeletal connectivity in a unified autoregressive framework. It is trained on OmniRig, a large-scale multi-category rigging dataset containing 79,499 meshes with skeleton and skinning annotations.
@inproceedings{sun2025armo, title = {{ARMO}: Autoregressive Rigging for Multi-Category Objects}, author = {Sun, Mingze and Mao, Shiwei and Chen, Keyi and Chen, Yurun and Lu, Shunlin and Wang, Jingbo and Dong, Junting and Huang, Ruqi}, booktitle = {Proceedings of the IEEE/CVF International Conference on Computer Vision}, pages = {7721--7730}, year = {2025}, month = oct, }
CVPR
DRiVE: Diffusion-based Rigging Empowers Generation of Versatile and Expressive Characters

Mingze Sun, Junhao Chen, Junting Dong, Yurun Chen, Xinyu Jiang, Shiwei Mao, Puhua Jiang, Jingbo Wang, Bo Dai, and Ruqi Huang

In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Jun 2025

Abs Bib HTML PDF Website

DRiVE generates and rigs expressive 3D Gaussian characters with complex garments and hair. Its GSDiff module predicts joints as spatial distributions, while the AnimeRig dataset supplies large-scale skeleton and skinning supervision.
@inproceedings{sun2025drive, title = {{DRiVE}: Diffusion-based Rigging Empowers Generation of Versatile and Expressive Characters}, author = {Sun, Mingze and Chen, Junhao and Dong, Junting and Chen, Yurun and Jiang, Xinyu and Mao, Shiwei and Jiang, Puhua and Wang, Jingbo and Dai, Bo and Huang, Ruqi}, booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition}, pages = {21170--21180}, year = {2025}, month = jun, }