About Dobb-E
Dobb·E is a cutting-edge productivity and automation tool designed to revolutionize the way robots learn and perform household tasks. By leveraging imitation learning, this open-source framework addresses the challenges facing current home robotics, providing a cost-effective and user-friendly solution for gathering demonstrations. At the heart of Dobb·E is the ingenious "Stick" tool, created from a $25 reacher-grabber stick, 3D printed parts, and an iPhone, enabling users to collect data seamlessly and efficiently. With access to the Homes of New York (HoNY) dataset, which comprises 13 hours of interactions across 22 different homes, Dobb·E is equipped with a diverse array of RGB and depth videos, along with action annotations for the gripper's 6D pose and opening angle. This wealth of data is crucial for training Dobb·E's representation learning model, known as Home Pretrained Representations (HPR). Built on the ResNet-34 architecture and utilizing self-supervised learning objectives, HPR initializes a robot policy capable of executing new tasks in various environments. One of the standout features of Dobb·E is its impressive performance. The framework has been shown to achieve an 81% average success rate in completing novel tasks within just 15 minutes, requiring only five minutes of collected data in a new home. This capability makes Dobb·E an ideal choice for anyone looking to enhance their home automation experience. Dobb·E offers a range of applications, from improving household chores to assisting elderly individuals or those with disabilities. The ease of use and affordability of the Stick tool democratizes access to advanced robotics, allowing a broader audience to engage with and benefit from this technology. Whether you're a robotics enthusiast, a researcher, or someone looking to automate everyday tasks, Dobb·E provides a versatile platform that can be tailored to meet your needs. Best of all, Dobb·E is entirely free, making it an accessible choice for individuals and organizations interested in pioneering the future of home robotics. Users can find extensive documentation, pre-trained models, and the underlying code on GitHub, ensuring that anyone can get started with Dobb·E without barriers. For those interested in the methodology and results, the open-access paper titled "On Bringing Robots Home" offers valuable insights into the framework's development and achievements. In summary, Dobb·E stands at the forefront of productivity and automation, transforming how robots learn and execute household tasks. Embrace the future of home robotics with Dobb·E and discover how this innovative tool can enhance your daily life. For more information, visit [Dobb·E's website](https://dobb-e.com/).
Key Features
- ✅ Empower robots to efficiently learn household tasks through innovative imitation learning technology.
- ✅ Achieve an impressive 81% average success rate in completing novel tasks within just 15 minutes of data collection.
- ✅ Utilize the affordable and user-friendly Stick tool, crafted from a simple reacher-grabber stick, to seamlessly gather task demonstrations.
- ✅ Access a rich dataset of 13 hours of interactions from 22 different homes, enhancing robot training with diverse real-world scenarios.
- ✅ Leverage the Home Pretrained Representations (HPR) model built on ResNet-34 architecture for robust and adaptable robot policies.
- ✅ Democratize access to advanced robotics with an entirely free, open-source framework that invites users to engage with cutting-edge technology.
- ✅ Support varied applications, from automating household chores to assisting individuals with disabilities, ensuring broader usability.
- ✅ Find comprehensive documentation, pre-trained models, and source code on GitHub to facilitate easy implementation and experimentation.
Pricing
Free to use
Rating & Reviews
3/5 stars based on 1 reviews