My “litter robot” is one of my favorite purchases, although it’s more like an appliance than a robot. Combined with an air purifier, my home never smells like cat litter.

My Roomba works ok, the mapping feature is neat but it still gets tripped up on small objects, is regularly defeated by cat hair, and doesn’t clean as effectively as a manual vacuum cleaning. I’m glad I have it.

My mopping Roomba is a hassle and I wish I didn’t purchase it.

I really want a textile folding robot but that seems like one of the more difficult robotics problems to solve generally. The dexterity and visual understanding needed is incredibly high.

Robots are hard. Not in the sexbot way, but mechanically difficult with more limitations than people imagine. The more powerful actuators are, e.g. for moving some weirdo humanoid robot, the heavier it is and the more power it will consume. There's no magical way around it. Mechanics also tend to break and will require maintenance, which sounds like a huge hassle. And some rolling "smart" screen loaded with sensors won't set foot, or wheel, in my home. The whole idea is f-ing creepy.

I’ve wanted to build a little companion robot that has internet connectivity through wifi and an android phone/tablet as a “face” combined with LLMs and all the other crazy AI tools launched lately to just be able to move around in a room and hold basic kid-friendly conversations. I realize this is probably an iceberg of a project with many complexities, but it does seem like the tooling/infrastructure is available now for something like this

What things can do a robot in the house apart of vacuuming?

What about looking on these tasks through the lens of negentropy? As in how much effort is needed to go from disorder to order.

One can reasonably say the negentropy of collecting little pieces scattered on the ground into a bag is low, even simple draft of air can do most of it. More advanced manipulation of stuff like laundry needs much more of it. And it would help us understand Moravec's paradox.

Maybe it is possible to compare negentrpy required for various activities and focus the robotics r&d on those with lower negentropy first? To provide better optimization gradient than mere guessing.

>> Whoever builds the next great home robot won’t have done so in a vacuum.

Badum-tish.

A lot of old school robotics methods need to die. Older professors who refuse to get with the times are holding the field back. For example, the whole graph-based navigation stack is dumb. Lidar-based SLAM is dumb. The field needs to go all in on AI/ML for navigation, planning, reasoning, language, and everything else.