Cewu Lu

About

Cewu Lu is a prominent researcher specializing in robotic perception, 6D object pose estimation, and robotic grasping, with particular expertise in bridging computer vision and practical robotics applications. His work addresses some of the most challenging problems in enabling robots to perceive and interact with the physical world. Lu's most influential contributions include DenseFusion, a pioneering approach to 6D object pose estimation through iterative dense fusion of RGB-D data, and 6-PACK, a real-time category-level 6D pose tracker leveraging learned 3D keypoints (159 citations). His grasp detection research has been equally impactful, with works like "Graspness Discovery in Clutters" (124 citations) and "RGB Matters" (117 citations) advancing reliable 7-DoF robotic grasping using monocular RGBD inputs. He has also tackled challenging edge cases such as transparent object manipulation through the TransCG dataset (116 citations) and articulated object understanding via the AKB-48 knowledge base (64 citations). Lu's collaborative contribution to Open X-Embodiment (119 citations) demonstrates his engagement with large-scale, generalizable robotic learning — a frontier of modern AI. Collectively, his research has garnered hundreds of citations, establishing him as a leading voice in intelligent robotic manipulation and 3D scene understanding.

Research Focus

Key Achievements

Top Papers

1. 1
   6-PACK: Category-level 6D Pose Tracker with Anchor-Based Keypoints

   159 citations · 2020
2. 2
   Graspness Discovery in Clutters for Fast and Accurate Grasp Detection

   124 citations · 2021
3. 3
   Open X-Embodiment: Robotic Learning Datasets and RT-X Models : Open X-Embodiment Collaboration⁰

   119 citations · 2024
4. 4
   RGB Matters: Learning 7-DoF Grasp Poses on Monocular RGBD Images

   117 citations · 2021
5. 5
   TransCG: A Large-Scale Real-World Dataset for Transparent Object Depth Completion and a Grasping Baseline

   116 citations · 2022
6. 6
   Open X-Embodiment: Robotic Learning Datasets and RT-X Models

   101 citations · 2023
7. 7
   DenseFusion: 6D Object Pose Estimation by Iterative Dense Fusion

   93 citations · 2019
8. 8
   AKB-48: A Real-World Articulated Object Knowledge Base

   64 citations · 2022
9. 9
   Toward Real-World Category-Level Articulation Pose Estimation

   37 citations · 2022
10. 10
    Skeleton Merger: an Unsupervised Aligned Keypoint Detector

    35 citations · 2021

Key Collaborators

Contact & Links