Zhiyuan Zhang

Zhiyuan Zhang 张志远

I'm a first-year PhD student at Purdue University, advised by Professor Yu She. I am a recipient of the Dr. W. Dale & Mrs. Jeanne Compton Fellowship (2025-2026) and the IE Graduate Excellence Award (2025).

I received my Bachelor's and Master's degrees in Mechanical Engineering from Huazhong University of Science and Technology.

G. Scholar / Twitter / Github / LinkedIn

News

[April 2025] I was honored with the IE Graduate Excellence Award 🎉
[September 2025] I was honored to receive the Dr. W. Dale & Mrs. Jeanne Compton Fellowship 🎉
[May 2026] Canonical Policy was accepted for publication in IEEE Transactions on Robotics (T-RO) 🎉
[June 2026] CONTACT was accepted to IROS 2026 🎉
[June 2026] EquiBim was accepted to IROS 2026 🎉

Research

* indicates equal contribution.

My research aims to design learning algorithms for robotic agents, enabling them to perform everyday manipulation tasks with human-level proficiency. To this end, I am currently focusing on multimodal robot learning and world models. Specifically, my research explores:

Geometric robot learning and equivariant representations.
Multimodal perception with vision and touch.
World models for planning and decision making.

	Imagining the Sense of Touch: Touch-Informed Manipulation via Imagined Tactile Representations Zhiyuan Zhang, Adeesh Desai, Jyun-Chi Hu, Yosuke Saka, Quan Khanh Luu, Jiuzhou Lei, Davood Soleymanzadeh, Bihao Zhang, Minghui Zheng, Yu She Under Review, 2026 website / arXiv / video / code TacImag learns to predict task-relevant tactile representations from visual observations, enabling touch-informed manipulation at deployment without physical tactile sensors.
	ContactWorld: What Matters in Vision-Tactile World Models for Contact-Rich Manipulation Zhiyuan Zhang, Pokuang Zhou, Kaidi Zhang, Adeesh Desai, Temitope Amosa, Davood Soleymanzadeh, Jiuzhou Lei, Minghui Zheng, Yu She Under Review, 2026 website / arXiv / video / code ContactWorld reveals that representation structure matters, tactile benefits depend on cross-modal compatibility, and tactile sensing becomes increasingly valuable for long-horizon planning.
	EquiBim: Learning Symmetry-Equivariant Policy for Bimanual Manipulation Zhiyuan Zhang, Aditya Mohan, Seungho Han, Wan Shou, Dongyi Wang, Yu She IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2026 website / arXiv / video / code EquiBim enforces symmetry-equivariant behavior by ensuring that actions predicted from an observation and its mirrored counterpart transform consistently under reflection.
	CONTACT: CONtact-aware TACTile Learning for Robotic Disassembly Yosuke Saka, Jyun-Chi Hu, Adeesh Desai, Zhiyuan Zhang, Bihao Zhang, Quan Khanh Luu, Md Rakibul Islam Prince, Minghui Zheng, Yu She IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2026 website / arXiv / video / code CONTACT reveals when and why tactile sensing matters for robotic disassembly, highlighting the effectiveness of structured tactile force-field representations in contact-dominated tasks.
	EquiForm: Noise-Robust SE(3)-Equivariant Policy Learning from 3D Point Clouds Zhiyuan Zhang, Yu She Under Review, 2026 website / arXiv / video / code EquiForm enables noise-robust equivariant observation-to-action mappings by correcting noise-induced geometric distortions and aligning representations under SE(3) transformations.
	Canonical Policy: Learning Canonical 3D Representation for SE(3)-Equivariant Policy Zhiyuan Zhang, Zhengtong Xu, Jai Nanda Lakamsani, Yu She IEEE Transactions on Robotics (T-RO), 2026 website / arXiv / video / code Canonical Policy enables equivariant observation-to-action mappings by grouping both in-distribution and out-of-distribution point clouds to a canonical 3D representation.
	ManiFeel: Benchmarking and Understanding Visuotactile Manipulation Policy Learning Quan Khanh Luu, Pokuang Zhou, Zhengtong Xu, Zhiyuan Zhang, Qiang Qiu, Yu She Under Review*, 2025 website / arXiv / video / code ManiFeel is a reproducible and scalable simulation benchmark for studying supervised visuotactile policy learning.
	GelRoller: A Rolling Vision-based Tactile Sensor for Large Surface Reconstruction Using Self-Supervised Photometric Stereo Method Zhiyuan Zhang, Huan Ma, Yulin Zhou, Jingjing Ji, Hua Yang IEEE International Conference on Robotics and Automation (ICRA), 2024 paper / video / bibtex @INPROCEEDINGS{10610417, author={Zhang, Zhiyuan and Ma, Huan and Zhou, Yulin and Ji, Jingjing and Yang, Hua}, booktitle={2024 IEEE International Conference on Robotics and Automation (ICRA)}, title={GelRoller: A Rolling Vision-based Tactile Sensor for Large Surface Reconstruction Using Self-Supervised Photometric Stereo Method}, year={2024}, volume={}, number={}, pages={7961-7967}, keywords={Mechanical sensors;Surface reconstruction;Accuracy;Shape;Shape measurement;Tactile sensors;Lighting}, doi={10.1109/ICRA57147.2024.10610417}} Close By leveraging GelRoller, users can quickly sense large surfaces and reconstruct the 3D shape of the contact region using just a single input image.

Professional Service

Reviewer for International Journal of Robotics Research (IJRR)
Reviewer for IEEE Transactions on Robotics (T-RO)
Reviewer for IEEE Transactions on Automation Science and Engineering (T-ASE)
Reviewer for IEEE Robotics and Automation Letters (RA-L)
Reviewer for Conference on Robot Learning (CoRL)
Reviewer for IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

Awards

IE Graduate Excellence Award, Purdue University, 2025
Dr. W. Dale & Mrs. Jeanne Compton Fellowship, Purdue University, 2025
Digital Electronics Scholarship, Huazhong University of Science and Technology, 2023
Xiaomi Scholarship, Huazhong University of Science and Technology, 2020

Website template from Jon Barron's website.