Slashdot Mirror
Sony's AIBO robot Sold Out
An Anonymous Coward writes "Sony's AIBO entertainment
robot sold out in Japan
in 20 minutes. They sold 3000 units; that's more than
2 units sold per second!" Sony has been advertising this
thing on Slashdot actually so I've browsed the site pretty
extensively. I just wish I had the cash to fork out for
one of these puppies (rimshot). They look awesome, although
I think I want a version that can climb stairs...
Last Christmas, I wandered over to my favourite Christmas Lights area to check out the decorations. One of the people there had a little plastic robot that his son was playing with. He was a really nice guy, so I hung out around his house for a while, checking out the robot. It couldn't do much sophisticated - it could walk around, say what you told it to via the remote control, and flash a few lights. I can tell you that his son just ate it up.
I think the thing cost about $ 100. Aibo costs twenty times what the robot did, but it's capable of independent behaviour of some kind. It also looks like it's made out of considerably more solid materials.
I'd be a little wary of the value for money proposition. My understanding is that kids get bored with toys pretty fast, and it's an awfully expensive toy. When we get the kind of mania that sells out a toy in seconds, especially something this expensive, I think there are factors more than sheer merit driving the phenomenon. Obviously, Sony's marketing is brilliant, but I doubt this has the long-term potential a $2,000 toy should.
I echo the thoughts of a fellow slashdotter in this thread - I'd blow $2,000 on a comprehensive Lego Mindstorms set and see what my kid (if I had one) could create, instead of getting this pre-packaged temptation. I think it would create a lot more enjoyment.
D
----
. . . there's always the upcoming Yoda (R) Furby (R).
That depends. You're assuming that the "dog" would operate with some sort of intelligence not only *on par* with organic minds, but also *compatibly* with organic minds. If I were going to build an AI pet (AI people would be different), then I would keep out a lot of the features. Sure, it would be neat if it was fully functional, but it would be much more convenient to make a dog where you didn't need to feel guilty about neglecting the thing, because it wouldn't mind. It's a very tricky topic, but I think that conventional ethics would only apply if its brain functioned like our does, or like those of dogs do.
I'm speaking more of general robot intelligence here, and ignoring your Turing test comment; that wouldn't apply if its brain functioned differently.
But see, now I feel guilty about contemplating neglecting a robot dog, even though it wouldn't care--very difficult subject matter indeed.
Well, a few thousand units really isn't a lot for a commercial product like this. I can imagine quite a few kinds of people willing to pay:
1. A Japanese businessman lives in a small apartment in Tokyo. His kid wants a pet. He goes for one that doesn't make any messes and won't wake up the neighbors. Also you can substitute "aged parent" for "kid".
2. Somebody who just loves electronic gadgetry but is mega-rich and mega-busy, and doesn't have the time to build his own robot. I mean, once you're worth multi-millions (and so many geeks are nowadays), you simply don't have to bother with accounting for the thousands here and there.
How would you use the Turing test on a robotic dog, anyway?
Consider that the Turing test basically states that to pass, it must be able to mimic a human so well that it is impossible to tell the machine from the genuine article. In the area of intelligence, it's been agreed upon that this means using language as intelligently as a human would (which would be possible via an IRC-like mechanism).
So, how would you apply the Turing test to a robotic dog? Would you send up up against a team of dog experts (veterinarians, dog trainers, etc.) to see if it were indistinguishable from a real dog? We're far away from that one happenning.
No, it runs a proprietary OS, designed for real-time operation. Actually, the doggie has a number of parallel processors running -
* A CognaChrome vision system - runs on some version of 680x0 chip, probably a specialized
embedded one. The last CognaChrome I played with
ran on a custom 68020 board - but Sony must have
done something extra special with it.
What is a CognaChrome? It's a fairly simple but very effective vision system originally from MIT. Basically, it thresholds the image to extract blobs of color, and then it returns the position of the biggest blob. I played around with an earlier version CognaChrome, it's relatively easy to work with and VERY FAST. The first version only recognized one color. I believe the doggies recognize 3 or 4, which largely depends on the processor speed. But hell, it's probably the only really workable real-time computer vision system right now!
* One or more motion control processors, doing nothing but servo control. This puppy has 3 servos per leg (to get 3 degrees of freedom), two to hold its head up (or down, or to the side), and one for the tail. That's a lot of PID loops to keep track of.
* a central processor to run the "intelligence" and command the rest of the system. Last time I heard, it was a custom version of the MIPS
R4000 chip (same as PlayStation?).
The doggies are very programmable, unless Sony put some deliberate obfuscations in. I've seen the "alpha-doggie" up close, it actually had an
Ethernet card hooked up to it so you could telnet
to it and tweak parameters. Of course, they probably removed that for the commercial version.
It's too bad I'm not doing robotics anymore, if I stayed at my last job, I probably would have had one!
mAx
--> Any fool can criticize - and many do --
Sony has a monopoly on the market, whatever that market is, so they can charge as they will...
The minute someone else has another robotic pet on the market, the price will drop, the manufacturing rate will increase, etc.
Personally, I want a robotic Pikachu!
But I'm weird.
-AS
-AS
*Pikachu*
Sony isn't the typical /. advertisor, and the banner add came soon after the original /. article on the bot, so I wonder if the slashdot effect itself atracts advertisers:
"We sudenly got a lot of hits from some place called slashdot, that had a story about out product. Hey, the sell banner adds, and their readers are obviously interested in our product, so lets advertise there!"
That'd be nice.