Some further reading on the LLM-and-robotics aspect in addition to your link:

https://qwen.ai/blog?id=qwen-robotsuite

https://drivingbench.com/

I'm not sure if I would consider tech demos like this state of the art. They look impressive and they do show promise but there's a huge gap between an impressive demo and a usable product. This goes for Google Gemini as well.