Dobb·E

About Dobb·E

Dobb·E is a groundbreaking open-source framework aimed at revolutionizing home robotics through rapid learning of household tasks. Using just five minutes of user demonstrations and a uniquely designed tool, Dobb·E trains robots to adapt and efficiently handle various domestic chores, catering to tech enthusiasts and researchers.

Dobb·E offers a free access model, encouraging users to explore its powerful features without financial commitment. For advanced users, premium options provide exclusive tools and resources, enhancing performance and training capabilities. Users benefit from exceptional training speed and efficiency, making Dobb·E a must-have resource for robot developers.

Dobb·E features a user-friendly interface designed for seamless navigation and efficient task management. Its layout allows users to easily access tools and documentation, ensuring a smooth experience when training robots. This intuitive design enhances user engagement and simplifies the complex process of robotic training.

How Dobb·E works

Users interact with Dobb·E by onboarding through a simplified setup process utilizing the demonstration collection tool, known as the Stick. After collecting five minutes of task demonstrations, users engage with the platform’s intuitive interface to upload data, access training modules, and monitor progress, leading to efficient robotic task execution.

Key Features for Dobb·E

Rapid Task Learning

Dobb·E’s rapid task learning feature allows robots to master household tasks in just 20 minutes. By leveraging user demonstrations, the platform efficiently adapts to new environments, providing a versatile solution for home robotics enthusiasts looking to enhance their robotic assistance capabilities.

The Stick Tool

The Stick is a unique demonstration collection device central to Dobb·E’s functionality. Built from affordable components, it enables users to efficiently gather data for teaching robots. This innovative tool simplifies the data collection process, empowering users to effectively share their tasks with robotic systems.

Homes of New York Dataset

The Homes of New York (HoNY) dataset is a comprehensive collection of 13 hours of interaction data from diverse household environments. Used to train Dobb·E’s models, this dataset enhances the platform’s adaptability and reliability in learning from real-world scenarios, ensuring success in varying household tasks.