A Tokyo robotics company shut down; I reviewed my own test logs
I compared the simulation environment and real-world deployment paths by actually running through them. I'm not surprised that companies like Integral AI went under. The hardest part of robotics has never been making a robot perform an action, but ensuring that action repeats stably a thousand times a day in the real world.
Physix Frontier