Yunrui Lian | 连允睿
|
CV |
Email |
Google Scholar |
| LinkedIn | Github | RedNote |
|
I am a second-year M.S. student at IIIS, Tsinghua University, advised by Prof. Li Yi. My research is supported by GALBOT.
Prior to that, I received my Bachelor's degree from Fudan University. I also spent time as a research intern at Carnegie Mellon University.
My research lies at the intersection of robotics and machine learning.
I am currently interested in developing algorithms that enable robots to learn complex, dynamic, contact-rich, and natural human skills, and to generalize them across diverse tasks and environments.
My long-term goal is to help build the general-purpose humanoid robot perception-decision-action system.
|
- [09/2026] LATENT received the Best Entertainment and Amusement Paper Award (main conference) at IROS 2026.
- [06/2026] LATENT was accepted to IROS 2026.
- [06/2026] CAT was accepted to RA-L 2026 (Transferred to ICRA 2027).
- [03/2026] We present LATENT, a method that constructs a latent action space from imperfect human motion data, enabling a humanoid robot to acquire diverse, natural, human-like tennis skills and sustain extended rallies with different human players in the real world.
- [02/2026] Humanoid-GPT was accepted to CVPR 2026.
- [01/2026] Any2Track was accepted to ICRA 2026.
|
Publications
* Equal contribution · First-author papers
|
|
Systematically Exploring the Capabilities of GPT-6 Astra as Embodied Policies
Galbot Team
arXiv, 2026
We systematically evaluate GPT-6 Astra as an embodied policy across 6 domains (gripper manipulation, dexterous manipulation, mobile manipulation, navigation, locomotion, and humanoid loco-manipulation), characterizing its strengths in task-level decisions and policy collaboration, alongside limitations in reliable physical control and inference efficiency.
webpage |
arXiv |
abstract |
bibtex |
code (manipulation)
GPT-6 Astra exhibits a remarkable ability to generate numerical robot actions, extending its role beyond high-level planning. To assess Astra’s capabilities as general-purpose embodied policies, we conduct comprehensive evaluations across six domains, examining direct control, cooperation with learned policies, and feedback-driven adaptation. In gripper manipulation, Astra can correct task targets and prepare contact conditions for subsequent policy execution; hybrid control with π0.5 achieves 48% success on the evaluated RoboDojo subset. In dexterous manipulation, hybrid control achieves 50% success in ten experience-guided DexJoCo trials, while direct in-hand control struggles to coordinate finger contacts. In mobile manipulation, hybrid control reaches 38.7% success on the evaluated RoboCasa365. In navigation, Astra leads our local comparisons, reaching 92% success on RxR instruction following and 82% on HM3D object search, although search incurs substantial detours. In locomotion, dense motion-reference generation remains unreliable: none of five sequential attempts on a single obstacle course reaches the goal, despite improvements in stability and forward progress. In humanoid loco-manipulation, Astra exceeds baseline methods on 13 of 30 HumanoidBench tasks with pretrained whole-body controllers. These findings reveal a gap between useful task decisions and reliable physical control. Inference latency further constrains practical control: across 50 RoboDojo instances per condition, policy-assisted and direct control consume 624.8 million and 1.132 billion tokens. A 30-second locomotion run requires 250 model calls averaging 39.86 seconds each, with physics paused during inference.
@misc{galbotteam2026systematicallyexploringcapabilitiesgpt6,
title = {Systematically Exploring the Capabilities of GPT-6 Astra as Embodied Policies},
author = {Galbot Team and Xuchuan Chen and Xiaoqian Cheng and Yu Deng and Lihe Ding and Shaocong Dong and Xiangjun Gao and Haozhe Jia and Zekai Li and Zhoujian Li and Yunrui Lian and Sikai Liang and Chenghuai Lin and Dairu Liu and Jiahang Liu and Qingtao Liu and Yuxuan Ma and Zekun Qi and Jiayi Su and He Wang and Ruochen Xu and Tianyu Xu and Xudong Xu and Zhe Xu and Mi Yan and Siming Yan and Li Yi and Ruixi Yu and Jinlu Zhang and Yintianrun Zhang and Zhikai Zhang and Zhizheng Zhang and Yixin Zheng and Weiyi Zhu},
year = {2026},
eprint = {2609.38537},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.38537},
}
|
|
|
PASSAGE: Scaling Scene-Aligned Motion Learning for Perceptive Humanoid Traversal in Cluttered Environments
Yuxuan Ma*, Zicheng Zeng*, Chunlin Peng*, Zhoujian Li, Zetong Zhao, Zhikai Zhang, Yunrui Lian, Han Xue, Sikai Liang, Weiyi Zhu, Mulin Chen, Chenghuai Lin, Jiayu Zeng, Yanwei An, Songan Zhang, Jiayuan Gu, Jilong Wang, Jingbo Wang, He Wang, Li Yi
arXiv, 2026
We present PASSAGE, a perception-conditioned planner–tracker framework for humanoid traversal. We scale up scene-aligned human motion data to hundreds of hours across thousands of cluttered scenes, enabling data-driven learning of human-like obstacle-avoidance skills.
webpage |
arXiv |
abstract |
bibtex |
code (coming soon)
Humanoid robots can step over, squeeze past, and duck under obstacles, but learning to select and coordinate these behaviors from onboard perception remains challenging. Many existing approaches rely on task-specific reinforcement-learning objectives or curated motion libraries, making broad behavioral coverage costly. We present PASSAGE, a perception-conditioned planner--tracker framework for humanoid traversal. Using virtual reality and inertial motion capture, we collect 100 h of scene-aligned human motion across 1,500 cluttered scenes. A conditional flow-matching planner generates short-horizon references from motion history, a local destination, and a robot-centric multi-layer elevation map, while a perceptive whole-body tracker executes them at 50 Hz with geometric feedback. Real-time chunking promotes inter-chunk consistency, and planner-side RL post-training under the frozen tracker further improves closed-loop performance. Without skill annotations or obstacle-specific policies, one planner--tracker pair selects and composes traversal behaviors across unseen geometries. In simulation, component ablations quantify the contribution of each stage. Across three independent training seeds, scaling captured data from 6 to 100 h increases mean contact-free success from 48.1% to 68.9% on held-out scenes, while the final model with validated scene augmentation reaches 70.3%. The fully onboard system integrates egocentric 3D LiDAR perception, online occupancy mapping, 6.25 Hz planning, and 50 Hz control on a Jetson AGX Orin; tests across 50 unseen physical layouts demonstrate traversal without prebuilt maps or offboard computation.
@misc{ma2026passagescalingscenealignedmotion,
title = {PASSAGE: Scaling Scene-Aligned Motion Learning for Perceptive Humanoid Traversal in Cluttered Environments},
author = {Yuxuan Ma and Zicheng Zeng and Chunlin Peng and Zhoujian Li and Zetong Zhao and Zhikai Zhang and Yunrui Lian and Han Xue and Sikai Liang and Weiyi Zhu and Mulin Chen and Chenghuai Lin and Jiayu Zeng and Yanwei An and Songan Zhang and Jiayuan Gu and Jilong Wang and Jingbo Wang and He Wang and Li Yi},
year = {2026},
eprint = {2609.18732},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2609.18732},
}
|
|
|
Learning Athletic Humanoid Tennis Skills from Imperfect Human Motion Data
Zhikai Zhang*, Haofei Lu*, Yunrui Lian*, Ziqing Chen, Yun Liu, Chenghuai Lin, Han Xue, Zicheng Zeng, Zekun Qi, Shaolin Zheng, Qing Luan, Jingbo Wang, Junliang Xing, He Wang, Li Yi
Best Entertainment and Amusement Paper Award (main conference)
IROS, 2026
We present LATENT, a method that constructs a latent action space from imperfect human motion data, enabling a humanoid robot to acquire diverse, natural, human-like tennis skills and sustain extended rallies with different human players in the real world.
webpage |
arXiv |
abstract |
bibtex |
code (LATENT)
Human athletes demonstrate versatile and highly-dynamic tennis skills to successfully conduct competitive rallies with a high-speed tennis ball.
However, reproducing such behaviors on humanoid robots is difficult, partially due to the lack of perfect humanoid action data or human kinematic motion data in tennis scenarios as reference.
In this work, we propose LATENT, a system that Learns Athletic humanoid TEnnis skills from imperfect human motioN daTa.
The imperfect human motion data consist only of motion fragments that capture the primitive skills used when playing tennis rather than precise and complete human-tennis motion sequences from real-world tennis matches, thereby significantly reducing the difficulty of data collection.
Our key insight is that, despite being imperfect, such quasi-realistic data still provide priors about human primitive skills in tennis scenarios.
With further correction and composition, we learn a humanoid policy that can consistently strike incoming balls under a wide range of conditions and return them to target locations, while preserving natural motion styles.
We also propose a series of designs for robust sim-to-real transfer and deploy our policy on the Unitree G1 humanoid robot. Our method achieves surprising results in the real world and can stably sustain multi-shot rallies with human players.
@misc{zhang2026learningathletichumanoidtennis,
title = {Learning Athletic Humanoid Tennis Skills from Imperfect Human Motion Data},
author = {Zhikai Zhang and Haofei Lu and Yunrui Lian and Ziqing Chen and Yun Liu and Chenghuai Lin and Han Xue and Zicheng Zeng and Zekun Qi and Shaolin Zheng and Qing Luan and Jingbo Wang and Junliang Xing and He Wang and Li Yi},
year = {2026},
eprint = {2603.12686},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2603.12686},
}
|
|
Humanoid-GPT: Scaling Data and Structure for Zero-Shot Motion Tracking
Zekun Qi*, Xuchuan Chen*, Dairu Liu*, Chenghuai Lin*, Yunrui Lian, Sikai Liang, Zhikai Zhang, Yu Guan, Jilong Wang, Wenyao Zhang, Xinqiang Yu, He Wang, Li Yi
CVPR, 2026
We present Humanoid-GPT, a GPT-style Transformer trained on a billion-scale motion corpus for whole-body control, achieving zero-shot generalization to unseen motions and control tasks.
webpage |
arXiv |
abstract |
bibtex |
code (Humanoid-GPT)
We introduce Humanoid-GPT, a GPT-style Transformer with causal attention trained on a billion-scale motion corpus for whole-body control.
Unlike prior shallow MLP trackers constrained by scarce data and an agility–generalization trade-off, Humanoid-GPT is pre-trained on a 2B-frame retargeted corpus that unifies all major mocap datasets with large-scale in-house recordings.
Scaling both data and model capacity yields a single generative Transformer that tracks highly dynamic behaviors while achieving unprecedented zero-shot generalization to unseen motions and control tasks.
Extensive experiments and scaling analyses show that our model establishes a new performance frontier, demonstrating robust zero-shot generalization to unseen tasks while simultaneously tracking highly dynamic and complex motions.
@article{humanoidgpt26,
title = {Humanoid-GPT: Scaling Data and Structure for Zero-Shot Motion Tracking},
author = {Qi, Zekun and Chen, Xuchuan and Liu, Dairu and Lin, Chenghuai and Lian, Yunrui and Liang, Sikai and Zhang, Zhikai and Guan, Yu and Wang, Jilong and Zhang, Wenyao and Yu, Xinqiang and Wang, He and Yi, Li},
journal = {arXiv preprint arXiv:2606.03985},
year = {2026}
}
|
|
|
Collision-Free Humanoid Traversal in Cluttered Indoor Scenes (CAT)
Han Xue*, Sikai Liang*, Zhikai Zhang*, Zicheng Zeng, Yun Liu, Yunrui Lian, Jilong Wang, Qingtao Liu, Xuesong Shi, Li Yi
RA-L, 2026 (Transferred to ICRA 2027)
We present Click and Traverse. It enables collision-free humanoid traversal in cluttered indoor scenes with Humanoid Potential Field (HumanoidPF), a general guidance for learning obstacle avoidance skills.
webpage |
arXiv |
abstract |
bibtex |
code (Click-and-Traverse)
We study the problem of collision-free humanoid traversal in cluttered indoor scenes, such as hurdling over objects scattered on the floor, crouching under low-hanging obstacles, or squeezing through narrow passages.
To achieve this goal, the humanoid needs to map its perception of surrounding obstacles with diverse spatial layouts and geometries to the corresponding traversal skills.
However, the lack of an effective representation that captures humanoid–obstacle relationships during collision avoidance makes directly learning such mappings difficult.
We therefore propose Humanoid Potential Field (HumanoidPF), which encodes these relationships as collision-free motion directions, significantly facilitating RL-based traversal skill learning.
We also find that HumanoidPF exhibits a surprisingly negligible sim-to-real gap as a perceptual representation.
To further enable generalizable traversal skills through diverse and challenging cluttered indoor scenes, we further propose a hybrid scene generation method, incorporating crops of realistic 3D indoor scenes and procedurally synthesized obstacles.
We successfully transfer our policy to the real world and develop a teleoperation system where users could command the humanoid to traverse in cluttered indoor scenes with just a single click. Extensive experiments are conducted in both simulation and the real world to validate the effectiveness of our method.
@misc{xue2026collisionfreehumanoidtraversalcluttered,
title = {Collision-Free Humanoid Traversal in Cluttered Indoor Scenes},
author = {Han Xue and Sikai Liang and Zhikai Zhang and Zicheng Zeng and Yun Liu and Yunrui Lian and Jilong Wang and Qingtao Liu and Xuesong Shi and Li Yi},
year = {2026},
eprint = {2601.16035},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2601.16035},
}
|
|
|
Track Any Motions under Any Disturbances (Any2Track)
Zhikai Zhang*, Jun Guo*, Chao Chen, Jilong Wang, Chenghuai Lin, Yunrui Lian, Han Xue, Zhenrong Wang, Maoqi Liu, Jiangran Lyu, Huaping Liu, He Wang, Li Yi
ICRA, 2026
We present Any2Track. It learns to track multiple highly-dynamic, contact-rich motions (over 40 hours in total) under any disturbances (terrain, external forces, payloads) by introducing a dynamics world model for policy fine-tuning.
webpage |
arXiv |
abstract |
bibtex |
code (OpenTrack)
A foundational humanoid motion tracker is expected to be able to track diverse, highly dynamic, and contact-rich motions.
More importantly, it needs to operate stably in real-world scenarios against various dynamics disturbances, including terrains, external forces, and physical property changes for general practical use.
To achieve this goal, we propose Any2Track (Track Any motions under Any disturbances), a two-stage RL framework to track various motions under multiple disturbances in the real world.
Any2Track reformulates dynamics adaptability as an additional capability on top of basic action execution and consists of two key components: AnyTracker and AnyAdapter.
AnyTracker is a general motion tracker with a series of careful designs to track various motions within a single policy.
AnyAdapter is a history-informed adaptation module that endows the tracker with online dynamics adaptability to overcome the sim2real gap and multiple real-world disturbances.
We deploy Any2Track on Unitree G1 hardware and achieve a successful sim2real transfer in a zero-shot manner. Any2Track performs exceptionally well in tracking various motions under multiple real-world disturbances.
@misc{zhang2025trackmotionsdisturbances,
title = {Track Any Motions under Any Disturbances},
author = {Zhikai Zhang and Jun Guo and Chao Chen and Jilong Wang and Chenghuai Lin and Yunrui Lian and Han Xue and Zhenrong Wang and Maoqi Liu and Jiangran Lyu and Huaping Liu and He Wang and Li Yi},
year = {2025},
eprint = {2509.13833},
archivePrefix = {arXiv},
primaryClass = {cs.RO},
url = {https://arxiv.org/abs/2509.13833},
}
|
🏆 2025, Outstanding Graduate, Fudan University
🏆 2022, National Scholarship (Top 0.2% nationwide)
|
|