I imagine the goal here was to find a simple tasks to test how well the robots can handle it. The main purpose for humanoid robots is for use in the environments designed for humans. That’s really the only place you need them after all. Pressing buttons and taking measurements might not be exciting, but if it is a job that a human did before then it’s saving somebody from doing something really boring that’s still necessary.
I think the idea of using robots as a drop in replacement for human workers is pretty sensible in general. Unitree is already making them for only $16k a pop, this makes it very cheap to automate a lot of manual labor. Another nice aspect of such robots is that they’re versatile, so you can have them do many different jobs as the need arises. The cost of making a single robot to do a bunch of work humans are doing is far lower than to redesign entire systems to be automated.
That strikes me as weird. Wouldn’t it be easier to replace the manual switches with an automated system instead?
I imagine the goal here was to find a simple tasks to test how well the robots can handle it. The main purpose for humanoid robots is for use in the environments designed for humans. That’s really the only place you need them after all. Pressing buttons and taking measurements might not be exciting, but if it is a job that a human did before then it’s saving somebody from doing something really boring that’s still necessary.
I think the idea of using robots as a drop in replacement for human workers is pretty sensible in general. Unitree is already making them for only $16k a pop, this makes it very cheap to automate a lot of manual labor. Another nice aspect of such robots is that they’re versatile, so you can have them do many different jobs as the need arises. The cost of making a single robot to do a bunch of work humans are doing is far lower than to redesign entire systems to be automated.