- Ph.D. Student @ Department of Computer Science, The University of Hong Kong, advised by Prof. Ping Luo at MMLab@HKU
- Co-Founder @ Lumina Embodied AI
- Research Intern @ Tencent Robotics X
Education
Bachelor of Engineering
GPA 3.97 / 4.00. Ranked #1 out of 42.
Received the National Scholarship (2023), Zhejiang Governmental Scholarship (2022), and Zhejiang University Scholarship - First Prize (2023).
Visiting Student Researcher
Hybrid Robotics Group, advised by Prof. Koushil Sreenath.
Research Interests
I work on Embodied AI, with a focus on robot manipulation. Much of my recent research is about scalable infrastructure for robot learning. That covers generating large-scale simulation data and digital assets, and building unified benchmarks and evaluation standards for generalist manipulation policies.
I am also interested in embodied foundation models and vision-language-action (VLA) models that connect reasoning and visual grounding with dexterous control, and in reinforcement learning for manipulation and sim-to-real transfer.
If you’re interested in related topics and would like to collaborate, feel free to reach out! You can find my email in the hyperlink on the right panel.
Publications
Scalable Data & Benchmarks for Robot Manipulation
Benchmarks & Policy Evaluation
- XPolicyLab: A Unified Standard and Open Ecosystem for Robot Policy Evaluation and Deployment, Best Tool Paper Award. [arXiv]
- An Unexpected Robot Policy: Early Evaluations of GPT-6 Astra on RoboDojo and Beyond. [arXiv]
- RMBench: Memory-Dependent Robotic Manipulation Benchmark with Insights into Policy Design. [arXiv]
- RoboDojo: A Unified Sim-and-Real Benchmark for Comprehensive Evaluation of Generalist Robot Manipulation Policies. [arXiv]
- Benchmarking Generalizable Bimanual Manipulation: RoboTwin Dual-Arm Collaboration Challenge at CVPR 2025 MEIS Workshop. [arXiv]
- Advances and Innovations in the Multi-Agent Robotic System (MARS) Challenge. [arXiv]
Embodied Foundation Models & Visual Grounding
- RxBrain: Embodied Cognition Foundation Model with Joint Language-Visual Reasoning and Imagination (Core Contributor). [arXiv]
- GroundingPI: A Grounding Foundation Model towards Physical Intelligence with Visual Primitives. [arXiv]
- GroundAnything: Reconciling Parallel Decoding with Precise Visual Grounding at Flash Speed. [arXiv]
- Efficient Alignment of Unconditioned Action Prior for Language-conditioned Pick and Place in Clutter.
- Towards Trustworthy Embodied Intelligence: A Systems Framework and Graded Trustworthiness Levels. [arXiv]
Multimodal Models & Autonomous Driving
- VidUEU-Agent: A Data-Curation Agent for Multimodal Understanding, Editing, and Unified Tasks.
- What Generation Knows: Unlocking Fine-Grained Understanding in Unified Multimodal Models.
- IMUG-Bench: Benchmarking Unified Multimodal Models on Interleaved Understanding and Generation.
- Cove-DLM: A Continuous Embedding Diffusion Vision-Language Model.
- Drive-KD: Multi-Teacher Distillation for VLMs in Autonomous Driving.
- Drive-P2D: A Progressive Perception-to-Decision Benchmark for VLMs in Autonomous Driving.
Experience
Research Intern @ Tencent Robotics X
In-context reinforcement learning for cross-embodiment and object adaptation in in-hand manipulation. Core contributor to Hy-Embodied-RxBrain-1.0; contributor to Hy-Embodied-0.5-VLA.
Intern @ Lightwheel AI
Led a team developing RL-based manipulation skills. Contributed to an IsaacLab-based robot learning framework covering data collection, multi-task multi-node RL training, and RL fine-tuning of VLA models.
Visiting Researcher @ UC Berkeley Hybrid Robotics Group
Efficient reinforcement learning for humanoid manipulation, using 3D Gaussian Splatting rendering to improve sim-to-real transfer.
Research Intern @ Huawei Noah's Ark Lab & ZJU AIM Group
Robust metric depth estimation for autonomous driving built on Metric3D, and outdoor scene reconstruction improving FrozenRecon with a more accurate depth-filling algorithm.
Undergraduate Researcher @ ZJU Robotics Lab
Tightly-coupled visual-inertial-wheel odometry for resource-constrained robots, with a consistency proof via observability analysis, and filter-based distributed multi-robot SLAM.
Selected Honors
- Hong Kong PhD Fellowship (2025)
- HKU Presidential Scholarship (2025)
- National Scholarship (2023), National Encouragement Scholarship (2024)
- Best Poster Presentation Award, HKU Summer Research Program (2023)
- Zhejiang University Scholarship - First Prize, top 1% (2023)
- Zhejiang Governmental Scholarship (2022)
- CKC Pilot Scholarship, three-time recipient, top 3% (2022, 2023, 2024)
- Xiaomi Scholarship (2024), Liyue Venture Capital Fund Scholarship (2023)
Services
Co-Founder @ Lumina Embodied AI
- Building a community that helps embodied AI researchers, engineers, and builders find and collaborate with each other
- Organizing talks from academia and industry, themed workshops, and community events across cities
- Connecting talent and opportunities in the embodied AI field
Co-Organizer @ TriFusion Workshop, SIGGRAPH Asia 2025
- TriFusion: Towards Embodied Intelligence Across Humans, Avatars, and Humanoid Robotics
Teaching Assistant @ CS, HKU
Teaching Assistant @ Chu Kochen Honors College, ZJU
- Calculus (H), Linear Algebra (H), Fundamentals of Programming
Textbook Contributor @ Linear Algebra Left Undone (线性代数：未竟之美)
- Open-source lecture notes and exercises supplementing the linear algebra courses of ZJU’s Turing Class
Volunteer Teacher @ Gansu, Hunan & Qinghai (2021.10-2022.8)
- Taught mathematics and physics at middle and high schools in Gansu, Hunan, and Qinghai
- Led a volunteer teaching team at Huangjinao Middle School, Huaihua, Hunan