yeah, to be honest, robots specifically are not interesting to me. well, or more to the point, it is not widely understood that a lot of "thinking" that computes for physical motion happens right nearby the joints and muscles doing the motion. the reason is latency, and the compute is small, and distributed. most robot control systems have a central brain that controls everything, and especially for military use cases, or just generally, this means that wherever that brain is, and all the wires connected to it, are a kill switch. where little computers all across the body keep working so long as the properly designed power distribution works. or in other words, distributing the batteries as well as the brains would make the robot more: resilient to damage, faster at reacting to stimuli, and more able to recover when something bad happens.
but i would call this something well beyond my intended scope, even if it demonstrates the same principles of small, distributed, semantic aware computing. maybe in the future. it's just out of scope for the initial work, which is aimed at computers that keep running for 100 years.
