Sorry for the vague subject line, this is a bit off-topic:
I've been thinking for some time about the state of consumer robotics. Clearly robotics is common in industrial applications, but almost non-existent in the home. Except for Roomba and Scooba, almost nobody uses robots on a daily basis. Somewhere in the future I think this will cease to be true, and robots will be just as common and comfortable as PCs are today.
This parallel is the basis of the question I want to pose. AFAIU, personal computing was similarly non-existent in the 70's, and the PC explosion can largely be traced back to Apple computer and two guys named Steve working in a garage lab. (If my understanding of history is incorrect, please fill me in. I'm only 23, so this is just what I've read.)
So here's the question: what are the odds that the personal robotics revolution will be spurred by somebody building a killer-app machine in their garage? Is it even valid in the first place to say that there are similarities between personal computers and personal robotics?
One major difference I see is that computer technology is consistently being miniaturized, so that as the tech gets more advanced we use fewer materials to produce CPUs, RAM, ASICs, etc. Robots, on the other hand, will probably always require a significant amount of materials for the casing, mechanics, actuators, etc.
Also, I'm not sure if anybody has a clear idea of a what a personal robot would do just yet. Did anybody know in 1980 what a personal computer would be used for? I'm not sure of that answer, but it seems like we have to develop a list of uses for a personal robot if anybody is going to see the appeal in buying one.
Supposing that this robot would be priced the same way PCs original were priced, we have a ceiling of maybe $4000-5000 on the total price, although closer to $2000 would make more sense in today's world. So what would a robot need to do do earn a spot in the home budget?
Well the two most popular home robots right now can vacuum and mop, so those are features that need to be put in. A clever idea would be to decouple the cleaning unit from the drive train and logic unit, perhaps putting a vacuum attachment in a trailer that the robot can pull around and exchange for a mop attachment when it needs to clean hard floors.
Of course, it should also able to charge itself and otherwise run completely free of human intervention for as many as several months at a time. So it needs to be able to find its base station, but it also needs to be able to find any attachments it might need to use. I really think this calls for better navigation. Right now, Roomba just randomly zig-zags around the floor, but it has no idea when its cleaned the room because there's no methodology. A personal robot (PR) needs to be more systematic. (Actually, this indoor navigation problem is one I'm most interested in working on.)
It probably needs to be able to traverse stairs, because we can't expect people to pay $2000 per floor of their house. Cleaning the stairs is probably out of the question, but going up and down them isn't out of reach.
What else could it do? Maybe it can dust objects within its reach? Interface with home computers/security system/X-10 in order to help complete its objectives? (turn off lights when nobody is around, disable a motion sensor if it wants to clean the kitchen at 3am, upgrade its own firmware, report status, etc.) Pick up empty cans and throw them in the trash?
I think we all understand that taking even an extremely simple task like is nearly overwhelmingly complex to design and build into a robot, so what I'm asking is what are the low-hanging fruit? What will a PR look like when people open the box? Is it possible for one of us in this group to build the prototype in our garage? Or is it just simply too complex/require too much capital for a small group of independents to manage? If it is possible, what are going to be the assistive technologies that make it happen? Standardization is probably one of the keys, as well as putting a little more horsepower into the embedded computers we're using. Battery technology really needs a revolution too. Better actuators (series elastic actuators look promising). What else?
This is something I've been thinking a lot about because I like building things just for the enjoyment of problem-solving and having something to show for when you're done working, but I always have commercial applications in the back of my mind too. Is it too nuts to even think about?