The Future of Robotics, Part 3: When will we have servant droids?

In our three part series on the future of robotics, we've been trying to figure out just what it would take to build C-3PO for real. First we asked James Kuffner, a Carnegie Mellon professor who is working to build a robotic mind that may be as smart as a protocol droid's. The we grilled Matt Denton, and animatronic expert about what the big hurdles involved. Lastly, we're putting the question to Steve Norris, one of the editors of Robot magazine, whose marvelous gallery of homemade 'bots would put a Jawa sandcrawler to shame.

If there's anyone who could put C-3PO together, Norris is our man. Click Continue to see his take on the golden 'bot wonder.


Robot-Roundtable-Steve-Norris.jpgSteve Norris, homemade 'bot builder extraordinaire, also works as a full-time software architect when he's not tinkering with his gallery of high-end hobby robots. He writes some of the most complex articles on the subject for Robot magazine.

Norris is no stranger to DVICE. We first discovered him thanks to his clever robotic footstool, and paid him a second visit to check out his cool "Stonehenge" robotic arm clock.

DVICE: When can I get a C-3PO of my own?
Norris: Well, there are a few big hurdles. He's like the ultimate goal and I think he's probably been an inspiration to a lot of people in the robotics world, too. Like upright walking — it can take a tremendous amount of processing power just to walk. It's like all the hard problems are concentrated in building a robot. Also, obviously, his ability to reason.

In both of those aspects he's incredibly sophisticated. The AI part of him — being able to understand, follow orders, reason — and he's incredibly sophisticated when it comes to his mobility. There's very little you can compare him to today. I could see you doing parts of it

How close has your work brought you to a 'bot resembling a protocol droid?
I actually worked in AI back in the late '80s/early '90s, and there were a couple of different approaches, like C-3PO being able to understand what you're saying and then respond. We took a top-down approach to AI. We tried to emulate what the brain was doing, or at least the functions of the brain. That approach just didn't work.

There were a lot of people doing the opposite. "Let's simulate the neuron and work our way on up," [they'd say]. That also, again, worked out for simpler problems but it didn't scale up. So the idea is that it's probably something in between the two approaches.

What I find interesting is back in the '80s everyone thought in 10 years — in the '90s — we'd have a HAL 9000, that these problems would be an easy thing to lick.

Steve-Norris-homemade-robots.jpgSome of Steve Norris's robots: an autonomous footstool, an robot arm that tells time, and Huey, the color-chasing 'bot

What existing robot would you consider to be C-3PO's closest relation?
[The Honda Asimo is mobile], but it's not really reasoning. It's just preprogrammed sequences. It's not like it figured the layout of the Honda facility. But it's about as close as we've gotten so far. Certainly from a mechanical standpoint it's the closest we've come. They've made incredible strides when it comes to walking robots. [editor: pun not intended, we're pretty sure.]

MIT is doing lots of research in social robots, the idea of a robot interacting with a person. They started a while back (Qismo?). MIT was presenting a lot of sociable robots — one was a talking head, but it's supposed to help you stay on your diet or something. They're worried about all of the social aspects — the facial expressions, the eye movement — but it's almost like they're putting the cart before the donkey. I mean, it can't even move yet.

They're both progressing and maybe eventually they'll meet.

Would you count your own homemade robots as social machines?
No. I mean they really don't recognize your moods, nor do they actually convey them. All of the other robots are very non-sociable. They can basically follow you and stuff like that, so you can have them available when you need them. The only thing they recognize about me is that I'm at about 98.6°F and follow that thing.

[My robots] are kind of emulating something that you could become more emotionally attached to. Although if you can become emotionally attached to a footstool, I don't know. I wouldn't consider it sociable, but it does certainly evoke a response when they see robostool.

Steve-Norris-favorite-robots.jpgWhat robot — any robot at all — do you find most inspiring?
HAL 9000, from 2001: A Space Odyssey, although he's probably not really in the world of robots because he's a fixed machine. If I were actually to pick a robot — a mobile robot — the first real robot that I really remember is from Lost in Space; it's technically called the B9. The fact that it was very sociable, first of all. It could interact with its fellow members on the Jupiter 2. It was humorous as well as sarcastic. There was an attitude there I really liked about it, and its high level of intelligence. And there's just something endearing about him. Although from a mobility standpoint he probably can't do too much. Short arms. Robby is a close second, again because there's something endearing about him, his ability to interact. I like that sarcasm.