As a sibling comment implies though, there's also danger from it being stupid while unsupervised. For example, I'd be very nervous having it do something autonomously in my kitchen for fear of it burning down my house by accident.

From a different robot (Boston Dynamics' new Atlas) - the system moves at a "reasonable" speed. But watch at 1m20s in this video[1]. You can see it bump and then move VERY quickly -- with speed that would certainly damage something, or hurt someone.

[1] https://www.youtube.com/watch?v=F_7IPm7f1vI

Especially if holding a knife or something sharp.

or if you're old, injured, groggy from medication, distracted by something/someone else, blind, deaf or any number of things.

it's easy to take your able body for granted, but reality comes to meet all of us eventually.

They are designed to penetrate Holtzman shields, surely.

Are you saying it cannot move faster than they because of some kind of governor?

That is how I would design it. It is common in safety critical PLC systems to have 1 or more separate safety PLCs that try to prevent bad things from happening.

Although in a SIL safety system the dangerous events are identified and extremely thoroughly characterized as part of system design.

There cannot be a safety system of this type for a generalist platform like a humanoid robot. It's possibility space is just too high.

I think the safety governor in this case would have to be a neural network that is at least as complex as the robots network, if not more so.

Which begs the question: what system checks that one for safety?

Limiting max force applied CAN be can be characterized for this robot.

A governor, the firmware in the motor controllers, something like that. Certainly not the neural network though.