Rodney Brooks: Robotics In The Next 30 Years
Writing by Evan Ackerman on Wednesday, 15 of September , 2010 at 12:24 am
For Discover Magazine’s 30th anniversary, they’re posting a series of predictions from eminent scientists about what’s going to happen over the next 30 years. One of these scientists is Rodney Brooks, a professor of robotics at MIT and CTO of iRobot, and he’s got some interesting things to say (besides the all too familiar “robots right now are like computers in the 80s”):
One of the great things about the Roomba robot vacuum cleaner, which my company iRobot designed, is that it’s too cheap not to be autonomous. Military robots right now are too expensive to be autonomous—you can’t afford to have them screw up. If the Roomba misses a spot, no big deal, it can find it later. So there will be a lot more robot autonomy, but surprisingly it will start out at the low end. It will trickle up to the high end over time.
Too cheap not to be autonomous… The ‘trickle up’ idea for robotics isn’t something I’ve heard people talk about much. The conventional way of thinking is that expensive and complex robots with expensive and complex sensors will provide the origins of autonomy, and then as the hardware gets cheaper and more accessible, robots offering the same autonomous capabilities will also get cheaper and more accessible. After all, this is what happens with computers. Brooks is right, though, in that to some extent, the more expensive a robot is, the less likely we are to trust it entirely to itself. In order for true autonomy to trickle up from the bottom, however, we’re going to have to overcome the hardware limitations and start getting access to more technology like the $25 SLAM system in the Neato XV-11.
Cars will certainly be more robotic. There will be many more robots in our houses, in our hospitals, in our factories, and in the military. We don’t have armed robot soldiers yet, but if we did, a robot could afford to shoot second, where a person could not. A robot doesn’t care whether its life is at risk. One can imagine robots getting a lot better at detecting what’s dangerous and what’s not dangerous: where there’s a gun and where there isn’t one. They might actually become more moral than we are.
I know that this is a somewhat contentious subject, but I tend to agree with Brooks here, although I’m not sure about the morality bit… Plenty of people definitely don’t agree, but I think the idea of being able to shoot second with no risk is an important advantage that robots have and humans don’t.
[ Discover ]
Comments (7)
Category: General
- Add this post to
- Del.icio.us -
- Digg
Comment by brianlmerritt
Made Wednesday, 15 of September , 2010 at 12:36 am
I can’t believe they lost my input!
ps – great input from Mr Brooks!
Comment by Joey1058
Made Wednesday, 15 of September , 2010 at 10:46 am
Autonomy from the bottom up? Can you see all these robotics clubs scrambling to be the first to claim to be the first?
Comment by dex drako
Made Wednesday, 15 of September , 2010 at 10:37 pm
toy know I can see robot being more moral then humans for the same reasonly they’ll be faster, stronger and smarter then us.
WE MADE THEM TO BE SO.
robots are only what we make them to be and so if we teach the first generation of ture A.I. tobe more moral them we are any offspring they make will be more so. like any child it will be what the parents help make it.
Comment by eripsa
Made Saturday, 18 of September , 2010 at 3:53 pm
This has been Brook’s basic position since his antirepresentationalist, evolutionary-inspired arguments almost 20 years ago: if you want to develop intelligent machines, develop really simple, insect-like machines then build up from that basic architecture.
Brooks view seems just as controversial now as then, but the interesting thing to me is that on Brook’s view, there already exists a wide variety of autonomous robots. Autonomy isn’t some far off AI dream. It is just a description of the the degrees of freedom the machine has relative to some particular task; so, e.g., relative to vacuuming, the roomba can be considered autonomous. So “autonomy” isn’t a particularly high bar to meet, and it doesn’t necessarily implicate all the ethical considerations (like whether the robot is responsible for its actions) that is usually associated with the ethicist’s understanding of “autonomy”. See Haselager [2005] for more discussion on the differences between the roboticist’s and philosopher’s understanding of autonomy.
I love this blog!
Comment by Evan Ackerman
Made Saturday, 18 of September , 2010 at 11:43 pm
@eripsa Glad you like the blog! This goes back to the roots of the terminology: what you choose to call autonomous and what you choose to call a robot. I’m sure a lot of people would argue (me among them, at times) that a robot without autonomy isn’t actually a robot, but of course as you say, it’s relative to the task that the robot is designed to fulfill. I’m still curious as to why the conventional approach to high level autonomy is top down, when the other way seems to make more sense, even if it is slower.
Anyway, I found a PDF of that paper, and I’ll see if I can write a post on it next week.
Comment by eripsa
Made Sunday, 19 of September , 2010 at 12:20 pm
The top down assumption is probably a result of thinking that intelligence or autonomy is a uniquely human phenomenon, or at least paradigmatically human, so that understanding human intelligence and autonomy is the key to understanding those phenomenon generally. The idea is that the autonomy or robots (or animals, or whatever) is just a variation on the autonomy of fully rational persons, so the first step is to figure out how humans do it.
Even if the bottom up approach is slower (and I’m not sure it is), at least we have empirical evidence that it works; after all, thats how all intelligence on the planet evolved, from simpler parts.
Again, this is a horse that Brooks has been beating since the 80′s, to the chagrin of the anthropocentric top-downers. In his most famous [1991] paper, he says:
“We will never understand how to decompose human level intelligence until we’ve had a lot of practice with simpler level intelligences.”
No one who thinks that intelligence/autonomy are unique to homo sapiens will find this line satisfying at all, even if it is true.
Comment by cabin rentals
Made Monday, 21 of November , 2011 at 7:29 pm
I wish to express appreciation to the writer for this wonderful post.
