Carding 4 Carders
Professional
DeepMind and Robots: a game-changing combo!
One of the main challenges in robotics is the need for individual training of machine learning models for each robot, task, and environment. A new project developed by Google DeepMind and 33 other research institutions offers a solution to this problem. The goal is to create a universal AI system that can work with various physical robots and perform many tasks.
Pannag Sanketi, Senior Software Engineer at Google Robotics, noted: "Robots do a great job with specialized tasks, but they don't adapt well to new conditions. Usually, you need to train a separate model for each task, robot, and environment."
The Open-X Embodiment project introduces two key components: a data set containing information about different types of robots, and a family of models that can transfer skills for a wide range of tasks. These models were tested in robotics laboratories on different types of robots and showed excellent results compared to conventional training methods.
The Open X-Embodiment project was inspired by large language models (LLMs), which when trained on large, shared datasets can compare or even outperform smaller models trained on highly specialized datasets. The researchers found that this principle also applies to robotics.
The RT-1-X model was tested on various tasks in five research laboratories on five common robots. Compared to the specialized models developed for each robot, the RT-1-X showed a 50% higher success rate in tasks such as moving objects and opening doors.
Sergey Levin, an associate professor at UC Berkeley and co-author of the paper, wrote: "Such models usually 'never' work on the first attempt, but this one worked."
Researchers are considering integrating current advances with innovations in DeepMind's RoboCat model. In addition, the team presented the Open X-Embodiment dataset and an abbreviated version of the RT-1-X model to the public.
Sanketi concluded, "We hope that providing data and models will accelerate research. The future of robotics depends on learning robots from each other and, more importantly, allowing researchers to learn from each other."
One of the main challenges in robotics is the need for individual training of machine learning models for each robot, task, and environment. A new project developed by Google DeepMind and 33 other research institutions offers a solution to this problem. The goal is to create a universal AI system that can work with various physical robots and perform many tasks.
Pannag Sanketi, Senior Software Engineer at Google Robotics, noted: "Robots do a great job with specialized tasks, but they don't adapt well to new conditions. Usually, you need to train a separate model for each task, robot, and environment."
The Open-X Embodiment project introduces two key components: a data set containing information about different types of robots, and a family of models that can transfer skills for a wide range of tasks. These models were tested in robotics laboratories on different types of robots and showed excellent results compared to conventional training methods.
The Open X-Embodiment project was inspired by large language models (LLMs), which when trained on large, shared datasets can compare or even outperform smaller models trained on highly specialized datasets. The researchers found that this principle also applies to robotics.
The RT-1-X model was tested on various tasks in five research laboratories on five common robots. Compared to the specialized models developed for each robot, the RT-1-X showed a 50% higher success rate in tasks such as moving objects and opening doors.
Sergey Levin, an associate professor at UC Berkeley and co-author of the paper, wrote: "Such models usually 'never' work on the first attempt, but this one worked."
Researchers are considering integrating current advances with innovations in DeepMind's RoboCat model. In addition, the team presented the Open X-Embodiment dataset and an abbreviated version of the RT-1-X model to the public.
Sanketi concluded, "We hope that providing data and models will accelerate research. The future of robotics depends on learning robots from each other and, more importantly, allowing researchers to learn from each other."
