"In January 2024, Figure signed its first commercial agreement with BMW to deploy its humanoid robot in the German carmaker’s production facility in Spartanburg, South Carolina.
Now, the California-based robotics firm has released a video showcasing its 01 humanoid robot executing its first job by participating in the vehicle assembly process."
Is so slow. But guess cheaper. No breaks, no benefits, no pizza party.
It’s also going to keep getting better and faster at it, too.
And this is presumably an early model, and it wouldn’t surprise me if they specifically want it to appear careful for the demo video.
Yes. I don’t think enough people realise the significances of this fact. Unlike us, AI will never peak; it will always relentlessly get better.
Figure announced during its deployment in January that the robots would undergo training for twelve to twenty-four months. After this training period, they will be integrated into the facility with the precise skills required for each task.
That’s a pretty long training period. I wonder if any of the training data is reusable if the assembly process changes slightly?
Figure aims to create a global model that can manage billion-unit humanoid robots. The company points out that there are about 10 million unsafe or undesirable jobs in the US alone.
Not sure how I feel about 10M workers being dismissed over time… the robot itself is really impressive regardless
I have a feeling when BMW gets the maintenance and technician bills for this thing, they’ll realize that maybe meatbags aren’t so bad
From what I understand this is the first time training of the system. It’s brand new. Its going to take a while.
Some that would be analogous would be those arm robots. At the start they were very specific and I don’t think very flexible at all. But when I looked at them going into a factory they are very general purpose machines. The manufacturers would come out and install them and make changes easily. They were also talking about how they would train people onsite to make some alterations, e.g. new box sizes.
Seeing as its using neural networks, they are meant to be general purpose machines right? I’m sure they must be more flexible than other types of systems.
