Papers
21
Total Citations
756
H-Index
11
About
Hanbo Zhang is a leading researcher in robotic grasping and manipulation, with a focus on enabling robots to perceive and interact with complex, cluttered environments. His work centers on deep learning for grasp detection, visual reasoning, and human-robot interaction. Zhang’s major contributions include developing the Fully Convolutional Grasp Detection Network with Oriented Anchor Box (2018, 226 citations), which introduced a real-time approach for predicting multiple grasping poses from RGB images, and REGNet (2021, 82 citations), an end-to-end network for grasp detection in point clouds. He advanced the field by addressing object stacking and clutter through multi-task convolutional neural networks (2019, 76 citations) and the INVIGORATE system (2021, 47 citations), which combines visual grounding and natural language to grasp specified objects in clutter. Zhang also created the REGRAD dataset (2022, 43 citations) for safe, object-specific grasping. His recent work on vision-language foundation models as robot imitators (2023, 19 citations) showcases his ongoing innovation. With over 700 total citations, Zhang’s research has significantly improved robotic autonomy in unstructured environments.
Research Focus
Key Achievements
Top Papers
- 1Fully Convolutional Grasp Detection Network with Oriented Anchor Box226 citations · 2018
- 2Visual Manipulation Relationship Network for Autonomous Robotics82 citations · 2018
- 3
- 4
- 5A Real-Time Robotic Grasping Approach With Oriented Anchor Box70 citations · 2019
- 6INVIGORATE: Interactive Visual Grounding and Grasping in Clutter47 citations · 2021
- 7
- 8
- 9Visual manipulation relationship recognition in object-stacking scenes22 citations · 2020
- 10Vision-Language Foundation Models as Effective Robot Imitators19 citations · 2023