Richard Doan

University of California, Berkeley

Papers

2

Total Citations

1,439

H-Index

2

About

Richard Doan is a leading researcher in robotic manipulation and deep learning for grasping, best known for his pioneering work on the Dex-Net project. His primary research areas include robot grasp planning, synthetic data generation, and the application of deep learning to physical interaction tasks. Doan’s major contribution is the development of Dex-Net 2.0, a framework that trains deep neural networks to plan robust grasps using only synthetic point clouds and analytic grasp metrics. By generating a massive dataset of 6.7 million synthetic examples from thousands of 3D models, he demonstrated that robots could learn effective grasping policies without costly real-world data collection. This work, cited over 1,400 times, has become a cornerstone of data-efficient robot learning. Doan’s impact is evident in the widespread adoption of his methods for sim-to-real transfer, significantly reducing the time and expense required to deploy robotic grasping systems. His achievements have advanced the field of robotic manipulation, making it more accessible and scalable for industrial and service applications.

Research Focus

Key Achievements

2
H-Index
2
Papers
1,439
Total Citations
720
Avg Citations/Paper
🏆 Most Cited Paper
Dex-Net 2.0: Deep Learning to Plan Robust Grasps with Synthetic Point Clouds and Analytic Grasp Metrics
1,162 citations · 2017
📈 Most Prolific Year: 2017 (2 Papers)
🤝 Key Collaborators: 9
🏛 Institutions: University of California, Berkeley

Top Papers

  1. 1
  2. 2

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 13 days ago