Español
Português
Việt Nam
日本語
русский 
عربي 
Deutsch
한국어
Türkçe
Français
Italiano
March 13, 2024 - In a major breakthrough, OpenAI spin-off company Covariant announced that it has successfully built a new artificial intelligence model that enables robots to learn to perform a variety of tasks like humans. This development marks a move towards more flexible and autonomous robotics.
In the summer of 2021, OpenAI shut down its robotics team, citing a lack of training data that had hampered its progress. However, three early-stage research scientists at Covariant found a solution at their startup, founded in 2017. They used years of data from fleets of picking robots in warehouses around the world, as well as text and videos from the Internet, to create a new model called RFM-1. This model combines the reasoning power of a large language model with the physical flexibility of an advanced robot.
The capabilities of the RFM-1 are impressive. Users can guide the model to perform tasks using five different types of input including text, images, videos, robot instructions, and measurements. For example, a user could show the model an image of a trash can filled with sports equipment and tell it to pick up a bag of tennis balls. The robot then performs the corresponding task and adjusts as needed.
Although the model has "human-like" reasoning capabilities, there are still limitations. During the demonstration, it was discovered that when some new concepts are presented, the model may not be fully understood. This indicates that the model requires more training data and further improvements.
Covariant says it plans to roll out the model to customers in the coming months and hopes to continue improving its performance and efficiency in real-world environments. They deploy the models in environments such as warehouses for testing and interact with real-world instructions, objects, and environments.
As AI technology advances, so does competition among companies using AI to power robotic systems. While there are still some problems to be solved, Covariant is committed to continuously learning and improving their models so that the robots can better adapt to changing environments.
In the near future, one can expect to see wider applications of robotics in various fields, and as more data and improvements become available, robots' ability to learn and perform tasks will continue to improve.