Xuguang Lan

About

Xuguang Lan is a prominent robotics researcher whose work sits at the intersection of computer vision and autonomous robotic manipulation. His research focuses primarily on robotic grasping, visual perception, and human-robot interaction, with a particular emphasis on enabling robots to operate intelligently in complex, cluttered environments. Lan's most influential contribution, "Fully Convolutional Grasp Detection Network with Oriented Anchor Box" (2018), introduced a real-time deep learning framework for predicting robotic grasp poses from RGB images, accumulating over 226 citations and establishing him as a leading voice in vision-based grasping. He extended this work into 3D point cloud processing with REGNet (2021, 82 citations), tackling the challenge of reliable grasping from partial, noisy observations. A defining thread throughout his research is addressing multi-object scenes — his Visual Manipulation Relationship Network (2018, 82 citations) and the REGRAD dataset (2022) tackle object relationships and occlusion, problems that traditional grasp detection methods routinely fail to handle. Perhaps most ambitiously, his INVIGORATE system (2021) integrates natural language understanding with visual grounding and grasping in clutter, pointing toward truly interactive autonomous robots. With hundreds of citations across his portfolio, Lan's work meaningfully advances the field's capacity to deploy robots in real-world, unstructured settings.

Research Focus

Key Achievements

Top Papers

1. 1
   Fully Convolutional Grasp Detection Network with Oriented Anchor Box

   226 citations · 2018
2. 2
   Visual Manipulation Relationship Network for Autonomous Robotics

   82 citations · 2018
3. 3
   REGNet: REgion-based Grasp Network for End-to-end Grasp Detection in Point Clouds

   82 citations · 2021
4. 4
   A Multi-task Convolutional Neural Network for Autonomous Robotic Grasping in Object Stacking Scenes

   76 citations · 2019
5. 5
   A Real-Time Robotic Grasping Approach With Oriented Anchor Box

   70 citations · 2019
6. 6
   INVIGORATE: Interactive Visual Grounding and Grasping in Clutter

   47 citations · 2021
7. 7
   REGRAD: A Large-Scale Relational Grasp Dataset for Safe and Object-Specific Robotic Grasping in Clutter

   43 citations · 2022
8. 8
   Task-oriented Grasping in Object Stacking Scenes with CRF-based Semantic Model

   23 citations · 2019
9. 9
   Visual manipulation relationship recognition in object-stacking scenes

   22 citations · 2020
10. 10
    Fully Convolutional Grasp Detection Network with Oriented Anchor Box

    22 citations · 2018

Key Collaborators

Contact & Links