Justin Sullivan/Getty Images
Researchers at the University of California Berkeley have developed a framework that allows robots to teach each other without human intervention.
Historically, robots struggle to transfer skills between models, which means that they rely on human instructions to complete a task.
According to the University of California Berkeley, a new framework that’s made up of two modules, Ro-Aug, which makes demonstrations across various robots, and Vi-Aug, which simulates diverse viewpoints has made major strides in robotic learning.
The RoVi-Aug uses models to create visual demonstrations that allow robots to learn from a broader range of experiences, including different hardware types and camera angles.
The findings were released in a 25-page research study.
“The success of modern machine learning systems, particularly generative models, demonstrates impressive generalizability and motivated robotics researchers to explore how to achieve similar generalizability in robotics,” researchers Lawrence Chen and Chenfeng Xu, told Tech Xplore.
By including different training data, the RoVi-Aug makes it easier to transfer skills, reducing the need for retraining robotics, according to Tech Radar.
This development could usher in robotics eliminating millions of jobs in the upcoming years, with experts saying smarter robots will revolutionize industrial jobs.
