When we look at the historical development of artificial intelligence, we see models that worked under certain conditions in the past how slowly it develops and it turned out how expensive it was. Therefore, over time, when neural networks developed sufficiently, they replaced outdated systems and led to the emergence of large language models. A similar change may occur for robots.

Researchers from the Massachusetts Institute of Technology (MIT).introduced a method similar to the working logic of large language models, which can be used instead of robot development methods that can only work under certain conditions. With this method, robots are trained with data from different sources and data that they collect themselves. This makes it easier for them to learn a new job or adapt to new circumstances.

Inspired by language models

In developing this method, researchers ChatGPT He stated that they were inspired by great language models such as. Based on the fact that large language models can be trained with large-scale language data and then used with very little training, the researchers concluded that “Heterogeneous, pre-trained transformers” He developed a model called (HPT). This model allows data from different cameras and sensors to be processed together, making it easier for robots to learn new tasks.

Thanks to the new method, robots can do both more skillfully They can move and perform different tasks. Researchers will now investigate the development potential of this method. They explain their purpose as robot brains that can be downloaded and installed like downloading an application.