|
Stanford University won the Darpa Challenge last weekend. You can read more about that here. What I find interesting about all of this is that: - Last year, none of the entrants in the challenge completed the course. In fact, none of them got more than 7.5 miles.
- This year, no less than 5 entrants completed the entire 150 mile course.
- Of the 5 that finished, 4 actually completed the course in less than the allotted 10 hours.
- The Stanford team completed the course in just under 7 hours.
Now remember, these are robots. No human assistance allowed. And this isn't your average Sunday drive on the highway. These 'bots were traveling cross-country. Now granted, crossing the Mojave isn't exactly like driving over the Rocky Mountains, but you still have to navigate around things; you can't exactly plow over a cactus, for example, and there are other obstacles on the course that have been deliberately placed.For the contestants to have made such significant improvements in just about a year is, well, amazing. Now, granted, this is DARPA. This stuff is being designed with a military application in mind. However, it isoften the case that stuff developed for the military eventually finds its way into civilian applications. This is, I think, an application that can have definite civilian appeal, provided the cost can be reduced. After all, 7 Pentium-M chips, a radar system, a GPS tracker, 4 laser range finders, and 2 camera systems (one stereoscopic) aren't cheap. Leaving aside the subject of cars driving themselves, though, what this all suggests to me is some major leaps in Artificial Intelligence. This has been one of the Holy Grails of computing for the better part of the last 40 years. We have science fiction stories of machines that are self-aware (anybody remember 'HAL' from 'Space Odyssey'?). We might finally be getting close to science fiction becoming fact. What do you think? Tell us in the forums!
|
