Twenty years ago, the most advanced AI we had was SmarterChild on AOL Instant Messenger.
Now we're on the verge of robots having dreams that will help them learn beyond anything that was possible before.
From the Daily Caller:
NVIDIA says its latest general-purpose robot update can teach a humanoid machine using software entirely, bypassing the months of motion capture sessions most rivals still depend on.
The chipmaker unveiled 'GR00T-Dreams' at the Computex trade show in Taipei on Monday...
...pitching it as a shortcut that turns a single photo into short simulation videos and, from there, into the step-by-step instructions a robot needs to move. In effect, engineers feed the robot a 'dream' reel instead of hours of recorded factory walks, cutting development time and cost.
Using this platform, a robot is shown a demonstration by a human of "a single task in a single environment." After that, the robot is given similar prompts, and is instructed to visualize the prompts through the use of generative video, after which it selects the best generations to study.
The system "lets robots learn a huge variety of new actions with minimal manual captures." 👇
The next generation of robots closing lids on boxes like:
It's a clever idea! As the Daily Caller puts it, the AI "watches thousands of ultra-realistic practice rounds that only exist on a server, learns what succeeds, and steps onto the factory floor already knowing how to stack boxes or sort parts."
This simulated learning will allow armies of robots to be deployed quickly into job sites of the future without the need for intensive training from humans.
Very cool stuff! So long as it doesn't end with ... well, you know:
P.S. Now check out our latest video 👇
