There’s two parts to answering this question.
When will it be technologically possible?
When will it be legally permissible?
This is a question about two seperate but closely related fields - AI and humanoid robotics (androids).
To have General Intelligence AI needs to first master about 1,000 specific intelligences, and then come up with a master program to use them and their results generally. So far AI has largely mastered about 500 of the specific intelligences.
The all purpose generally useful domestic robot really requires General Intelligence AI, together with about 100 specific robot technologies (skin touch, vision, balance, grip, etc etc.). They’ve largely mastered perhaps the first 50 of these 100 specific robot technologies.
So when we have true General Intelligence AI, and have mastered all those things about touch and human interface skills, we will be able to build a true generally useful domestic service robot.
My best guess is:
by 2030 - about 10% chance.
by 2040 - about 55% chance.
by 2050 - about 90% chance.
But, like flying cars, driverless vehicles, and a bunch of other things, just because you can make something, doesn’t mean the law will allow you to use them.
A General Intelligence useful domestic robot has the following characteristics:
It can be hacked.
It can be stolen. (And like a car it can be reconfigured, reprogrammed, and resold).
It is intelligent (not like a dog, but like a human). So it will have access to understand everything it sees in your dwelling. Where does that sit with things like privacy, pornography, security, etc.
It requires appropriate laws to apply when a domestic robot steals, harms, breaks, things? Humans aren’t foolproof. We sometimes start fires, break things, etc. etc. Do you want an imperfect robot that’s only as good as a human, or will you only accept a perfect one that can’t harm the baby, start a kitchen fire, drop a vase, or slip on a greasy floor? And who / what is responsible if a autodrive car kills someone, or a domestic robot starts a kitchen fire?
It is a general purpose intelligent robot. The difference between your household robot and your battlefield robot is … not very much. So do you want terrorists to be able to purchase an army of domestic robots?
A competent General Intelligence general purpose domestic robot will be more intelligent that many / most of its owner / user / operators? Do you only want dumb robots? (See 4 above) Potential problem?
A general purpose General Intelligence robot would potentially replace all domestic staff (cleaners, cooks, baby sitters / nannies, window cleaners, gardeners, etc. etc.) This would be true for homes and businesses. The impact upon human employment will be HUGE.
Security, including national security. Does the USA want fifty million - ‘made in China’ robots in the USA? Does anyone else except China? Does anybody trust any robot made in any other country?
So there will be a bunch of issues to be navigated legally before General Intelligence domestic robots get to be released. And these will potentially take longer to resolve than it takes for technology to raise the issues. The first flying car was made in about 1985. But you still aren’t allowed to fly it. Where is it legal to drive an autodrive car?
In the meantime you have various specific intelligence and specific function devices from automated pool cleaners to automated vacuum cleaners, to smart fridges that order food, to domestic Google devices to control the temperature, the lighting, run a bath, or burglar secure the place.
Androids can now stand, jump, climb stairs, grip things, learn skills, and interface quite well with people in voice, facial expressions and body language.
There’s a way to go, but progress is frighteningly fast.
Hope this helps.