Sure, but that's hardly the only problem. The simple fact is most people don't have systems in their home that they want/need to interface with through an app or something. Here's a close to comprehensive list of the electronic things in my home: Several computers, Microwave, George Foreman Grill, Rice Cooker, Fridge, Thermostat, Toothbrush, Lights, Dish washer, Washer/Dryer

Mechanical engineering doesn't seem to be blocking point for any of those except maybe the washer/dryer and food related stuff. I could definitely see the opportunity to have a device that could portion out N grams of rice per person and then have it ready to eat when I get home, but for the rest, I don't see it as the stumbling block.

Machine vision is also a problem. One of the biggest I think. You could have a tiny tractor that grows you delicious veggies, but machine vision can't deal with plants.

Yeah, but if we had that we wouldn't be talking about "internet of things" we would be talking about "actual AI everywhere" which a whole next level of difficulty up.

Injection molding is the cheapest process on the planet, if you have a production run of >10k for a mass produced product. It is hard to find anything on Earth cheaper than an injection molded object.

That's a big damn if though.

I mean, for small runs.