As a mechanical engineer, I can assure you that it's probably farther out than 5-10 years before they're in most people's homes.The main reason, at least in my head, assuming you could knock out all of the actual scaling, convenience, and cost problems, is that if you want something unique and truly created by you, then you'd actually have to do the 3D modeling yourself, and that's not easy, nor cheap, nor time-forgiving! It might be shorter to order it and get it shipped Amazon Prime!
If I wanted to make a custom chess set in Solidworks right now
at work, it might take me a week to model exactly what I wanted, then I'd have to shoot it over to a very large, expensive machine that costs me a lot of money simply for the material/time, and I'd have to wait a day or two for it to actually create the pieces. Then, I'd have to walk over and throw all of them in a bath to dissolve the support structure that's necessary in helping the machine create the parts. If someone figures out how to speed it up, cheapen it up, shrink it down, and remove the 3D modeling cost, learning curve, and time to create all of the models in a decade, I'd be relatively impressed, but perhaps not stunned.
With all that said, I think the technology has enormous potential. HUGE, in fact. It's only a matter of time before you're downloading the solid models other people create off the web and popping out a sweet duck shaped coffee mug an hour or two later.