[she11c0de]

I wonder if it would be possible to setup a multi-agent reinforcement learning scenario where a few of those play tag over a complex terrain.

[rasz]

They would probably need 10x the technical woman power to keep them operational for longer than 10 minute stretches. The second article about the challenges articulates pretty well they are in the crawling stage of locomotion, constantly discovering and improving on hardware bottlenecks and weak points.