[SystemSafety] Elephants, dinosaurs and integrating the VLA model

[Les Chambers]

Hi All
Further to my claim, that standards development is likely to be overrun by 
elephants

Where once robots had to be specifically programmed with rules and precise 
movements, it seems things are changing. Google’s RT-2 robot integrates a large 
language model (LLM), trained with Internet data, to increase cognition. 
They’ve called it the vision-language-action (VLA) model. At the link below the 
robot was told to pick up the “extinct animal"; it grabbed the dinosaur. Google 
engineers plan to leverage improvements in LLMs to make their robots smarter. 
Their architecture allows them to swap out one LLM for another. Which makes you 
wonder what form an integration test would take?
Can anyone on this list refer me to where in the standards one can obtain 
guidance on how to engineer such a system safely? 

https://www.bit.ly/3OEOQjE

Cheers
Les
--
Les Chambers
les at chambers.com.au

https://www.chambers.com.au
https://www.systemsengineeringblog.com

+61 (0)412 648 992