Robots need a lot of training and studies to do tasks correctly and efficiently, with Google DeepMind's latest AI systems helping pave the way for machines to get well-versed in human tasks. The latest of what Google DeepMind has to offer is to aid these bots in navigating through humanlike actions, training them to be precise and accomplish the tasks at hand.
Google DeepMind's AI is Now a Robot Teacher for Specific Actions
Google DeepMind
In a new blog post from Google DeepMind, it introduced two new AI systems that are responsible for training robots, offering a new and improved system that usher in significant improvements for the machines.
There is the new ALOHA Unleashed which is the upgraded version of its ALOHA 2, and it is responsible for helping bots learn and perform two-handed multi-faceted tasks.
While ALOHA Unleashed is more on the practical side, DemoStart is meant for the more technical knowledge aspect of its training regimen. This specific system is responsible for running the simulations that could be applied to the "real-world" performance of the machines, centering on multi-fingered robots.
The tandem of ALOHA Unleashed and DemoStart uses human demonstrations as its source of knowledge, and then translates these "images to actions."
According to Google DeepMind's study, its AI systems recently taught a robot with two-arm actions, a.k.a. "bi-arm manipulation," making it possible to tie shoelaces. While this is known as a simple and everyday task for humans, robots have limited complexities for now, and Google DeepMind claimed that its previous bots could only pick up things before ALOHA Unleashed and DemoStart.
Its bots' newfound skills include cleaning the kitchen, inserting gears, hanging shirts, and going as far as to repair another bot through its new AI teachers.
AI Systems for Machine Learning
Developers, researchers, and scientists have long used artificial intelligence in powering robots and other machines to give them complexity and depth in accomplishing certain tasks or duties that make lives easier. Both these advanced systems have been somewhat synonymous for many years now, with AI systems teaching robots how to laugh at the right jokes and moments.
Because of the massive advancements in technology, programmers can now deploy AI to share their knowledge and experiences with other machines, connecting them under one network for easier applications. A previous study by the University of South Carolina (USC) shared that AI robots can teach other bots, being mutually beneficial as it also picks something up from their counterparts.
The wide and expansive knowledge of artificial intelligence was made more available for users to enjoy, but one of its renowned purposes is to share its knowledge with less advanced machines. Google DeepMind's new AI systems are now the masters of robots, using its latest developments to improve their skills in multiple complex tasks, including duties that require dexterity and finesse.
