| CARVIEW |
Hello, I'm Wei Yang
I am a Senior Research Scientist at the NVIDIA Seattle Robotics Lab. I received my Ph.D. in Electronic Engineering from the Chinese University of Hong Kong in 2018, under the supervision of Prof. Xiaogang Wang and co-supervision of Prof. Wanli Ouyang. During my doctoral studies, I was a visiting student at the Robotics Institute, Carnegie Mellon University, where I worked with Prof. Abhinav Gupta.
Research interests: Robotics Manipulation, Computer Vision, Embodied Intelligence.
News
- September, 2025: VT-Refine (bimanual assembly with visuo-tactile feedback) and Dexplore (scalable neural control for dexterous manipulation) are accepted to CoRL 2025.
- Jan, 2025: Our Scene Synthesizer paper is published in JOSS and the code is now publicly available.
- May, 2024: Our FoundationPose paper is accepted to CVPR 2024.
- April 1st, 2023: One paper is accepted to CVPR 2023.
- Oct, 2022: Two papers are accepted to CoRL 2022 and IROS 2022/R-AL.
- May, 2022: Two papers are accepted to ICRA 2022.
- November, 2021: Our 6D grasping (GA-DDPG) paper is accepted to CoRL 2021.
- June 2, 2021: Our H2R handover paper won the 2021 IEEE ICRA Best Paper Award on HRI.
- March 3, 2021: Our DexYCB hand grasping object benchmark is accepted to CVPR 2021.
- Feb 28, 2021: Our reactive handovers paper is accepted to ICRA 2021.
- July 1, 2020: Two papers are accepted to IROS 2020.
- Jan 21, 2020: One paper is accepted to ICRA 2020.
- Jan 7, 2019: One paper is accepted to ICLR 2019.
- Jan 2, 2019: I join the NVIDIA Seattle Robotics Lab as a Research Scientist.
- Feb 28, 2018: One paper is accepted to CVPR 2018.
- Oct 23, 2017: Our team won the 2nd places of the PoseTrack Challenge 2017 (Technical report | Leaderboard).
- Oct 15, 2017: I am currently a visiting scholar at CMU working with Prof. Abhinav Gupta (from Oct 15 2017 to April 30 2018).
- July 17, 2017: Two papers are accepted to ICCV 2017.
- April 5, 2017: The code for our CVPR 2017 paper has been made publicly available now.
- March 18, 2017: One paper is accepted to CVPR 2017.
- June 3, 2016: I have fullfilled the candidacy requirement and will be advanced to Ph.D. (post-candidacy) stage.
Publications
Conferences & Preprints
Dexplore: Scalable Neural Control for Dexterous Manipulation from Reference-Scoped Exploration
A scalable neural control framework for dexterous manipulation using reference-scoped exploration.
VT-Refine: Learning Bimanual Assembly with Visuo-Tactile Feedback via Simulation Fine-Tuning
Learning bimanual assembly tasks using visuo-tactile feedback through simulation fine-tuning.
Slot-Level Robotic Placement via Visual Imitation from Single Human Video
Teaching robots slot-level pick-place manipulation from a single human video.
Cosmos World Foundation Model Platform for Physical AI
A world foundation model platform to help developers build customized world models for Physical AI applications with video curation pipeline, pre-trained models, and post-training examples.
FoundationPose: Unified 6d pose estimation and tracking of novel objects
A foundation model for 6D pose estimation and tracking of novel objects.
SynH2R: Synthesizing Hand-Object Motions for Learning Human-to-Robot Handovers
Using synthetic data to learn human-to-robot handovers.
AnyTeleop: A General Vision-Based Dexterous Robot Arm-Hand Teleoperation System
A general vision-based teleoperation system for dexterous robot manipulation.
Learning Human-to-Robot Handovers from Point Clouds
Learning human-to-robot handovers directly from point cloud observations.
Learning Robust Real-World Dexterous Grasping Policies via Implicit Shape Augmentation
Using implicit shape augmentation to learn robust dexterous grasping policies.
Learning Perceptual Concepts by Bootstrapping from Human Queries
Learning perceptual concepts through human interaction.
Model Predictive Control for Fluid Human-to-Robot Handovers
Using MPC for smooth human-to-robot handovers.
HandoverSim: A Simulation Framework and Benchmark for Human-to-Robot Object Handovers
A simulation framework for human-to-robot handovers.
Goal-Auxiliary Actor-Critic for 6D Robotic Grasping with Point Clouds
Using goal-auxiliary learning for 6D robotic grasping.
DexYCB: A Benchmark for Capturing Hand Grasping of Objects
A benchmark dataset for hand-object interaction.
Reactive Human-to-Robot Handovers of Arbitrary Objects
A reactive approach for human-to-robot handovers.
Human Grasp Classification for Reactive Human-to-Robot Handovers
A grasp classification approach for reactive human-to-robot handovers.
Collaborative Interaction Models for Optimized Human-Robot Teamwork
Optimizing human-robot teamwork through collaborative interaction models.
DexPilot: Vision Based Teleoperation of Dexterous Robotic Hand-Arm System
A vision-based teleoperation system for dexterous manipulation.
Visual Semantic Navigation using Scene Priors
Using scene priors for visual semantic navigation.
3D Human Pose Estimation in the Wild by Adversarial Learning
Using adversarial learning for 3D human pose estimation in unconstrained environments.
Learning Feature Pyramids for Human Pose Estimation
A pyramid network architecture for human pose estimation.
Identity-Aware Textual-Visual Matching with Latent Co-attention
Using latent co-attention for identity-aware textual-visual matching.
Towards Multi-Person Pose Tracking: Bottom-up and Top-down Methods
Bottom-up and top-down methods for multi-person pose tracking.
Multi-Context Attention for Human Pose Estimation
Using multi-context attention mechanisms for human pose estimation.
End-to-End Learning of Deformable Mixture of Parts and Deep Convolutional Neural Networks for Human Pose Estimation
Learning deformable mixture of parts and CNNs end-to-end for pose estimation.
Multi-task Recurrent Neural Network for Immediacy Prediction
Using multi-task RNNs for immediacy prediction.
Clothing Co-Parsing by Joint Image Segmentation and Labeling
Joint image segmentation and labeling for clothing co-parsing.
Data-Driven Scene Understanding by Adaptive Exemplar Retrieval
Using adaptive exemplar retrieval for scene understanding.
Learning Contour-Fragment-based Shape Model with And-Or Tree Representation
Using And-Or tree representation for shape modeling.
Interactive CT image segmentation with online discriminative learning
Interactive medical image segmentation with online learning.