Chuer PanHey there! I am a PhD student at Stanford University working in robotics, advised by Prof. Shuran Song. 🌊 Last summer, I interned at Boston Dynamics, working on dexterous manipulation for the electric Atlas! Previously,I got my MS in Robotics from CMU Robotics Institute, where I was fortunate to work with David Held and Martial Hebert on object-centric 3D representations for precise manipulation. I earned my BASc in Engineering Science, majoring in Robotics from the University of Toronto 🍁. There I worked on NLP with Frank Rudzicz at the Vector Institute. Earlier, I worked with Eugenia Kumacheva to fabricate human micro blood vessels :) Before grad school, I worked on FPGA R&D at Altera (then acquired by Intel), motion planning and controls for autonomous vehicle at aUToronto. I co-founded a machine learning student club (UTMIST) during undergrad, which has since become one of the largest undergrad machine learning-focused communities in North America! 🩵 Email: chuerpan at stanford dot edu |
|
ResearchI develop robotic systems that learn high-precision, contact-rich manipulation from human demonstrations and generalize to novel visual and behavioral settings with minimal adaptation. |
In-the-Wild Compliant Manipulation with UMI-FTHojung Choi*, Yifan Hou*, Chuer Pan, Seongheon Hong, Austin Patel, Xiaomeng Xu, Mark Cutkosky, Shuran Song International Conference on Robotics and Automation (ICRA), 2026 paper / website / UMI-FT is a handheld data-collection platform that integrates compact, finger-level force/torque sensors with multimodal vision data to facilitate force-aware policy learning. This system enables the training of adaptive compliance policies that effectively regulate contact forces during complex manipulation tasks, such as wiping or skewering. |
|
UMI-on-Air: Embodiment-Aware Guidance for Embodiment-Agnostic Visuomotor PoliciesHarsh Gupta, Xiaofeng Guo, Huy Ha, Chuer Pan, Muqing Cao, Dongjae Lee, Sebastian Scherer, Shuran Song, Guanya Shi International Conference on Robotics and Automation (ICRA), 2026 paper / website / UMI-on-Air, a framework for embodiment-aware deployment of embodiment-agnostic manipulation policies. We propose Embodiment-Aware Diffusion Policy, which integrates gradient feedback from an embodiment-specific low-level controller into an embodiment-agnostic high-level diffusion policy’s sampling process, steering trajectory generation toward dynamically feasible solutions for the deployment robot at test time. |
|
One Demo is Worth a Thousand Trajectories: Action-View Augmentation for Visuomotor PoliciesChuer Pan, Litian Liang, Dominik Bauer, Eric Cousineau, Benjamin Burchfiel, Siyuan Feng, Shuran Song Conference on Robot Learning (CoRL), 2025 paper / website / 1001-Demos is an effective data augmentation framework for robot manipulation, that scales single fisheye, eye-in-hand, obstacle-free, human demonstrations into action-view consistent demos with visually realistic observation & physically feasible action trajectories. By leveraging a novel fisheye-adapted Gaussian Splatting formulation and trajectory optimization, it generates deployable smooth, collision-free demos from novel viewpoints with added obstacles — making robot manipulation policies more robust to unseen configurations and learn to plan around obstacles. |
|
Universal Manipulation Interface: In-The-Wild Robot Teaching Without In-The-Wild RobotCheng Chi*, Zhenjia Xu*, Chuer Pan, Eric Cousineau, Ben Burchfiel, Siyuan Feng, Russ Tedrake, Shuran Song Robotics: Science and Systems (RSS), 2024 ★ Best Systems Paper Award Finalist, RSS 2024 ★ paper / website / We propose UMI, a portable data collection and policy learning framework that allows direct skill transfer from in-the-wild human demonstrations to deployable robot policies, allowing zero-shot generalizable dynamic, bimanual, precise, and long-horizon robot policies. |
|
Open X-Embodiment: Robotic Learning Datasets and RT-X ModelsOpen X-Embodiment Collaboration International Conference on Robotics and Automation (ICRA), 2024 ★ Best Conference Paper Award, ICRA 2024 ★ paper / website / We introduce the largest robot learning dataset to date with 1M+ real robot trajectories, spanning 22 robot embodiments. We train large, transformer-based policies on the dataset and show that co-training with our diverse dataset substantially improves performance. |
|
TAX-Pose: Task-Specific Cross-Pose Estimation for Robot ManipulationChuer Pan*, Brian Okorn*, Harry Zhang*, Ben Eisner*, David Held Conference on Robot Learning (CoRL), 2022 paper / website / We propose a vision-based system that learns to estimate the task specific pose relationship (cross-pose) between pairs of interacting object using learned cross-object correspondences. We demonstrate that our method is able to learn from just 10 real point cloud demonstration with no pose annotations needed and generalize to novel instances within the trained object category. |
|
|
Deep Projective Rotation Estimation through Relative SupervisionBrian Okorn*, Chuer Pan*, Martial Herbert, David Held Conference on Robot Learning (CoRL), 2022 paper / website / We propose a new algorithm for self-supervised orientation estimation which utilizes Modified Rodrigues Parameters to stereographically project the closed manifold of SO(3) to the open manifold of 3D Euclidean space, which avoids the local optima common when naively applying relative self-supervision for object orientation estimation, allowing for faster convergence and lower rotational error on relative rotation estimation. |
|
Examining the rhetorical capacities of neural language modelsZining Zhu, Chuer Pan, Mohamed Abdalla, Frank Rudzicz Conference on Empirical Methods in Natural Language Processing (EMNLP) BlackboxNLP Workshop, 2020 paper / video / We propose a method that quantitatively evaluates the rhetorical capacities of neural language models (LMs) by evaluating their abilities to encode a set of linguistic features derived from Rhetorical Structure Theory (RST). Our experiments show that BERT-based LMs outperform other Transformer LMs such as GPT-2 and XLNet, revealing richer discourse knowledge in their intermediate layer representations. |
|
An exploration of the reflow technique for the fabrication of an in vitro microvascular system to study occlusive clotsYang Li, Chuer Pan, Yunfeng Li, Eugenia Kumacheva, Arun Ramachandran Biomedical Microdevices, 19(4), 1-16, 2017 paper / We introduce a reflow technique for fabrication of multi-level microchannel network with circular cross-section to systematically study the dissolution effects of thrombolytic drug on occlusive embolic clots in microvascular system. |
Teaching
|
|
Design and source code adapted from Leonid Keselman's Jekyll fork of Jon Barron's website |
