scripsit

Elias presents ... a worm!    Thoughts on family, philosophy,
and technology

Profile

Wednesday, January 03, 2007

Best Pleo video yet

Yet another article on the forthcoming baby dinosaur robot, this one has a good five-minute video without any monologue, just an example round of playing with a Pleo: waking him up, poking at him playfully, inducing him to peer over the edge of a table, then picking him up until he goes back to sleep. The user in the video is the inventor. The three-page article is interesting too, giving background on the Idahoan inventor, who use to be a mime before creating the consumer-sensation Furby (which looks almost as annoying as those singing-and-dancing snowmen advertised during the holidays) and earning himself $10 million.

To be released in March at $250, the Pleo seems an unlikely purchase for us, but it sure is tempting. I'm just fascinated by the fact that we are at the cusp of the age of robots, and this will go down as one of the first mass-market attempts at a life-like robot. But then I am not sure if I want the kids to be around a robot pet just yet. Would it be fun for more than a week, then become a very expensive toy mistake? At the other extreme, would Elias get so emotionally attached to it that he'd cry when it breaks? What if he breaks it himself, will he be devastated? And is it just wrong to love a robot -- or is it good to develop a sense of both what a robot is and isn't? And isn't having pets and toys which need gentle care while they are alive/functioning a good part of life? Et cetera.

For my own part, I want to know more about the AI ware to get a good idea of how complex and therefore interesting this thing really is. Sounds like it is programmable to some extent (or maybe you can't write scripts, as you can for Lego bots, you can just set the Pleo's personality parameters), but there's not much it can physically do besides sleep, purr, wiggle, walk, and respond to pokes and pets. Hmm, sounds like a cat, minus the barf and litterbox ... that's starting to sound like a pretty good investment. Maybe this should be the marketing line: It's cheaper than a year of cat food.

UPDATE: Bill Gates compares, in five pages, the robotics industry today to the software industry of 30 years ago. Carr takes a quick glance.

Gates confirms something for me: the biggest challenges in robotics are software problems. But I'm skeptical of his suggestion that robotics is especially in need of multi-threaded software. His example of a robot too busy adjusting its drive motor to notice an imminent cliff is silly -- even Office bloatware is rarely that unresponsive -- and his praise for Microsoft's "concurrency and coordination runtime" just sounds like an ad to me. (Microsoft is obviously playing catch-up in robotics to other software giants ... like Lego.) If we've learned anything in software, it's that task switching on a Von Neuman architecture is a really, really good way to achieve virtual simultaneity. Moderately complex multithreaded programs are not just "difficult" to write, they're just shy of impossible for humans to get bug-free.

Labels:

3 Comments:

  • At 11:58 AM , Blogger softwareNerd said...

    Our son got a simple robot when he was around 5 and was bored of it pretty soon. There seems to be a certain age below which kids aren't too interested in automated toys. For instance, upto some age (can't remember when, but sometime in preschool) our son would not use the battery-operated cars as they were meant to. Instead, he would "drive" them around by hand. The same for the first battery engine we bought him. He turned it off for the first year and used it just like his other Thomas stuff.

    We got a MindStorm NXT this XMas and he likes it (he's 8 now), but it's not such a great hit. I think he'll "grow into" the NXT system though.

    BTW: Bill Gates has an article about robots, in a recent "Scientific American".

     
  • At 12:18 PM , Blogger Brad Williams said...

    Thanks for the interesting info. I can certainly imagine a toddler having little patience for the very limited behaviors of automated toys.

    I just saw the Gates article too, thanks.

     
  • At 8:32 AM , Blogger softwareNerd said...

    In essence, what's required to get robotic software going is some level of standardization at a "BIOS and OS" level. Just as we can program to a keyboard, mouse, etc. someone needs to abstract out the devices: sensors and motors and provide an interface that allows people to discover device properties (from among a std. set for that device-type) and program to that standard API.

    I think Gates is on the right track about concurrency if one views concurrency not so much as a timing speed issue, but more of an issue of "programming metaphor". I think the type of programming metaphor that will end up being the best starting point is one where we have "Event Listener" for the various sensors, with the Listener events triggerring various responses. It might also be helpful to layer this over some type state-transition. SO, a programming language that makes it easy to declare states and state-transistions (either by the language, or by an add-on library) could be helpful.

     

Post a Comment

Subscribe to Post Comments [Atom]

<< Home