A new AI breakthrough allows this robot to learn numerous tasks just by watching humans. This is thanks to the efforts of Figure AI, an AI robotics firm that aims to bring a general-purpose humanoid to life.
Figure AI posted a video of its accomplishment on Jan. 7. In its official YouTube content, the tech firm said that its humanoid bot called Figure-01 was able to make coffee just by watching videos of humans doing it.
But, making coffee is just among the wide range of tasks that Figure-01 can do. Here are other details you need to know about this advanced AI-powered machine.
AI Breakthrough Allows Robot Learn Tasks by Watching Humans
According to Fox News' latest report, Figure-01 can do a lot of tasks just by watching videos of humans. These include interacting, talking, walking, etc.
Figure AI conducted a demonstration at its headquarters in San Francisco, California. During the demo, a human asked the AI-powered robot to make him a coffee.
"Sure, I can do that!" cheerfully replied the advanced machine, which then proceeded to the Keurig machine.
Before the demo, Figure-01 watched a video of a human making coffee using the same machine. Via its official YouTube video, Figure AI said that the machine trained for 10 hours before doing the demo.
Figure-01 was able to open the Keurig coffee maker's hatch successfully. After that, the AI-powered humanoid robot inserted a coffee pod and then closed the hatch.
To make the coffee, it pressed the button all by itself and waited for the espresso to be poured into the mug. It waited for the process to be completed and let the human grab the coffee mug himself.
For some people, this is not really a big thing. However, it is an incredible feat for robotics, especially after it was able to self-correct its mistakes.
Read Also : US Army to Weaponize Robot Dogs by Mounting Rifles, LMGs; Integrating AI Capabilities Part of Plan
What AI Did Figure Used?
Figure-01's demo is quite astounding since it was able to perform a task autonomously without any assistance from a human. Aside from this, it only needed a video as its reference to learn how to make coffee.
The tech firm said that they used an end-to-end artificial intelligence system with neural networks that take video in and trajectories out.
Experts said that the demo of Figure AI was both impressive and impactful since what Figure-01 learned was added to its library. This new autonomous action can be transferred to other robots with the same end-to-end AI system.
This means that if one robot learns, the other machines also learn, thanks to what researchers call "swarm learning."
Related Article : China Seeks To Build Advance Humanoid Robots by 2025, Issues Guidelines for Innovation System
