As a lifetime technophile primarily interested in computers and videogames, another area that has always intrigued me, but been generally hands-off due to the various barriers to entry, is home robotics. There were some delightful robots and kits in the early to mid-80's to go along with the personal computer boom. However, the personal robotics boom was short lived and ultimately a much smaller niche than even the fledgling personal computer business at the time, dooming them to the domain of the truly hardcore. Today, toy and personal robots and robot kits from the likes of Tomy and Heathkit are still very much in demand. Much more recently, Lego made a strong impact in the home robotics and hobbyist field with their Mindstorms technology and Radio Shack carries an aggressive line of kit robots and accessories. Bottom line, today hobbyist robotics is stronger than ever and more practical than ever, though is still awaiting that "killer app" to truly push it into the mainstream.
The author has yet to crack open his Lego Mindstorms kit form a few years back, but he'll get to it eventually!
With that said, what has come closest to filling the awesome potential of personal and useful home robotics talked up for years have been the Roomba Robotic Vacuums. Robots that mow our lawns, vacuum and scrub our floors and generally perform maintenance duties logically have the greatest chance for success, particularly if they're easy to maintain, do a good job at what they're supposed to and don't cost a lot. The products from Roomba fit that criteria. So, it's no surprise that Roomba has finally fully embraced the robot hobbyist and enthusiast - though perhaps on a slightly more commercial level - as noted in this article here, entitled Roomba Robotic Vacuum Attracts Modders. Where there's a robot platform, you can't keep the modders and extenders away for long, even if it's not something as seemingly clear-cut as Sony's dearly departed Aibo line.
Who knows, I'm moving to a new home soon, so I just may be able to convince the wife of the practical benefits of a floor cleaning robot that just so happens to be hackable...
I'm also a big fan of robots, though I haven't done much either in the way of building them. One of the professors who was applying for a new media position here at SCSU showed a really funny film from the early days of tech that showed a bulky looking robot doing everyday chores, such as answering the door, watering the plants, making food, and so on. I think it was called "Leave it to Robo" or some such, but I can't seem to find it at the moment. Anyway, one of the general conceptions from the 50s on seem to be that robots should look something vaguely human-like. Perhaps this is more comforting than a robot that looks more like an insect.
I probably saw the concept film you were talking about. Human-like robots are coming as soon as the flying cars are. Futurism and fantasy are very different from reality and practicality. I watch every robot documentary/special I can on channels like Discovery, TLC and The Science Channel. Many think making humans more human- or animal-like or personable is the answer, and they're developing sophisticated algorithms and physical mechansims to make that a reality. Honda's Asimo is a good example of a successful robot that walks like a human. Others are focused more on facial expressions and having the robots "read" body language (something humans sometimes don't do well themselves). Others are going the utility route like the Roomba company, mentioned in my blog post above, where function dictates form. In the short- to medium-term I think that naturally will lead to the greatest success and the most integration of robotics into everyday life. After that, I'm sure we'll have successful humanoid multi-function robots. I suppose as a housemate for the elderly, for instance, where humanoid-like robotics are actually being put to use in limited trials in Japan and the US, that "softening" of the functional form is necessary to its success.
What's troubling is that we still don't have computers that understand us well, so to expect that from an advanced robot - along with all the other things they would have to do from a physical standpoint - is just too much to ask. It's like natural machine speech - that's still not perfected, though that has been promised for some time. The same thing goes for computer AI. It's just not there, and perhaps too much effort is put into the "kitchen sink" approach rather than just trying to perfect more individual elements.