Consumer robotics strikes me as an engineering tar pit so deep it leads to hell. If full self driving is hard due to the long tail of unusual special cases, this is orders of magnitude worse.
Take FSD but multiply the number of actuators and degrees of freedom by at least 10, more like 100. Add a third dimension. Add direct physical interaction with complex objects. Add pets and children. Add toys on the floor. Add random furniture with non-standard dimensions. Add exposure to dust, dirt, water, grease, and who knows what else? Puke? Bleach? Dog pee?
Oh, and remove designated roads and standardized rules about how you're supposed to drive on those roads. There are no standards. Every home is arranged differently. People behave differently. Kids are nuts. The cat will climb on it. The dog may attack it. The pet rabbit will chew on any exposed cords.
We've all seen those Boston Dynamics robots. They're awesome but how durable would they be in those conditions? Would they last for years with day to day constant abuse in an environment like that?
From a pure engineering point of view (neglecting the human factor or cost) a home helper robot is almost definitely harder than building and operating a Mars base. We pretty much have all the core tech for that figured out: recycling atmosphere, splitting and making water, refining minerals, greenhouses, airlocks, and so on. As soon as we have Starship or another super heavy rocket that's reliable we could do it as long as someone was willing to write some huge checks.
And of course it's a totally untested market. We don't know how big it really is. Will people really be willing to pay thousands to tens of thousands for a home robot with significant limitations? Only about 25% of the market probably has the disposable income to afford these.
You'd have to go way up market first, but people up market can afford to just pay humans to do it.