Do you consider say a 8 armed spider robot to be humanoid, because I don't.
I see I needed to be clearer about my contention. Humanoid was a poor choice of words. Human-equivalent would be better.
In other words, yes, a robot with 8 arms and tripod legs would absolutely qualify. Though what seems to be under development right now is humanoid, deviations could well make sense.
The key point I'm making is that the "specialized single task automated robots" are not the end goal. If you really want an automated home, you need something that can interact in the same way a person can. We already have the single task automated ones; programmable coffee makers, remote controlled thermostats, etc. They exist.
Or what about a magnetic or suction wheel that can attach to any surface ( including walls and ceilings ) in your home? Without even getting into levitating robots or robots that can extend 5 meter long limbs... All of these very non-humanoid robots configurations would be able to reach and manipulate more spaces and places that you can do and much faster as well.
Well, my ceiling is rough because of the paint used. Walls too. So I can't use one of those. Levitation...kind of doesn't exist? Flight is very energy intensive, and so if we don't NEED it (which we don't, because houses are designed to be used by humans which cannot fly), we are better off not wasting our limited budget (both financial and energetic) on that capacity. Extensible limbs are absolutely something you might include though.
Some physical limitations also doesn't apply when things can be done with robotic speed and precision. I bet if you could move your arm, hand and fingers with 100% perfect micrometer precision and superhuman speeds + without human limits how joints can be bent then it would be a piece of cake to fold any piece of clothing with a single arm.
Solid maybe there. Though due to safety concerns you do not necessarily want to deploy a machine that is moving that fast in an environment that humans or other fragile stuff might get into. That said, why bother? Two limbs are enough to solve that problem, without requiring clever manipulations of physics, increasing safety risks, or applying excessive forces to clothing that may not be able to handle it. Like. How is this solution BETTER than just using two arms? Human equivalent robots are desirable because they can do anything we might want them to do, not necessarily because they are strictly optimal at doing any one task. But in this specific example....
You don't just have to be able to do it. You have to be able to do it safely and economically.
Designing a humanoid robot capable of independent operation that is safe around small children... is a task I would not trust to be solved in the software layer. Whereas for most individual household functions you could probably come up with a reasonable way to child-proof a machine that does just that and only that function. If nothing else, then simply by putting it well out of reach of those children.
Custom machines that need to be safe around children and need to be designed for every single task are what we currently have. The change comes when we can deploy soft-exoskeleton robots which plan while requiring a recovery maneuver be available at all times (current active areas of research in the robotics community). Yes, this is a challenge. No, it is not unsolvable (obviously: humans manage to not kill each other walking around).
Humanoid robots are almost the Platonic ideal of plug-and-play substitution using multirole hardware. Multirole platforms are almost always either inferior to or more expensive than a well designed suite of specialized hardware: When employed in one role, all the other roles it can theoretically fill are adding design constraints without adding value. And plug-and-play replacements are almost always inferior to overhauling the system surrounding the thing you're replacing (simply because you get to explore a much wider solution space that way, for very minor additional cost in design complexity).
Yes, this is why we want them. The whole point is to upgrade our lives comprehensively. We ALREADY HAVE individual automation. To get things to work seamlessly, you can either try to get the various appliances to somehow interface with each other (and I've yet to see any proposal for how that isn't effectively a human-equivalent robot...even just for doing the laundry) or you can deploy something that can interact with the appliances in the same way a human would.
So the balance of probability is that a humanoid robot is going to be comparatively bad and expensive, unless there is a specific reason it needs to be human-looking. The most likely cause being that it has to stand in for a human in interpersonal relations.
Human equivalent robots are hard and expensive; it's why we don't have them yet and you have to fold your laundry yourself. But if you don't want to fold your laundry yourself, you end up needing a robot that has human level versatility or needing to renovate your entire house. Since many people will want to not fold their laundry but also don't want to or cannot renovate their entire house, there will be a market for such devices if and when we get them working.
This is not a near future thing, you are absolutely right on that point. Human equivalent robots are HARD. But they are categorically not impossible (humans exist, clearly these problems can be solved). The question is, do we want them. And my contention is that yes, we actually do.
