Dobb-E

Dobb-E

Train household robots with imitation learning

thumb

What is Dobb-E?

Dobb-E is an open-source framework that trains household robots using imitation learning. With an impressive 81% success in new tasks, it simplifies robot training. Utilizing the Stick tool and the Homes of New York dataset, Dobb-E gathers demonstrations quickly. Access pre-trained models and documentation on GitHub. Explore more in the research paper, "On Bringing Robots Home," to see how Dobb-E transforms home robotics.

Top Features

  • Open-source framework for household robots
  • Imitation learning for task training
  • Achieves 81% success in new tasks
  • Gathers demonstrations quickly
  • Access to pre-trained models and documentation

FAQs

Promote Dobb-E

Dobb-E