Great question. It's a combination of economics - lower-wage humans do many tasks much better - and technology - precise manipulation of tools and navigation of novel environments is incredibly difficult. Any small seemingly trivial task actually has an insane amount of complexity to it.<p>Let's say I wanted a robot to take out my trash. It sounds simple but there are so many incredibly difficult tasks when you break it down, each with a near-infinite number of variations in different homes:<p>- First, learn where in the house it is, and how to get to it.
* Is it in a drawer? What kind of drawer, how to open it?
* Is it a plastic garbage bag in a bin, with a foot lever? In a drawer?
- How does the robot lift out a plastic bag, replace the plastic bag? We don't have the dexterity to do this yet
* What happens when the plastic bag gets caught slightly on a corner, or begins to rip?
- Let's say we pick up the plastic bag, now we need to move to a door that will take us out of the house
* Are there stairs, pets, children, other obstacles that could get in the way? Just this bullet point here could be harder than self-driving cars, which is far from solved
* How do we interact with the door to open it? Is it a round knob, is there a deadbolt to unlock, does the door swing outward or inwards?<p>...etc<p>This probably barely scratches the surface on all of the variability inside of a home, and yet a small kid can do all this without even thinking, while even a small subset of these problems probably requires billions of dollars to solve in a controlled/closed environment.<p>Maybe humanoid robot teleoperation + artificial intelligence will get us there, that's the pipe dream of a lot of these humanoid robot companies. But then they need to make money and out-compete some young/lower-skilled workers happy to do things for $10/hour. At which point one wonders how these companies will make money to justify the insane R&D needed for even the simplest of tasks. But hey the same story has played out in other industries where robots have outplaced low-skilled labor. The difference though is that these environments have been heavily controlled, i.e. the same few steps to assemble a widget, the same motions to clean the same type of object, etc.