The January issue of Servo Magazine has FIRST on the cover. Unfortunately, other than the cover, there isn’t any copy devoted to FIRST. There is however a ‘feature’ article written by your ob’d. srvnt., Dr. Bot. So go out and buy lots of copies so I can get rich from royalties. (Just kidding, haven’t even got paid for the article yet!) In more robot news, the January Wired has
the 50 best robots of all time article(Pretty good) They got most of my favorites, but left out a couple including “Robot” from “Lost in Space,” and Huey, Dewey and Louie from “Silent Running.”
They also have a in depth article about Stanley from the DARPA Challenge. Not sure about how technically correct the article is, especially about the role of vision versus lidar in the programming. But I am voting for Stanley for Time’s “Entity of the Year.”
I liked how Stanley’s simpler scheme performed just as well, if not better than Red’s system. The article made it sound like Stanley would have tore through the desert if a mandatory speed limit hadn’t been enforced.
Err, from what I read in a WIRED article… Stanley’s system wasn’t very much simpler. Instead of programming Stanley to drive, they programmed Stanley to learn how to drive. Very interesting read. Wired 14.01: Say Hello to Stanley
I think that Stanley’s code is simpler in terms of less “if then” type conditionals, (generally speaking of course), as in constantly referencing set code in order to check what’s happening. The code becomes more complicated, but less complex (I think thats right…) by utilizing an algorithm that creates it’s own method of driving, i.e., it drives for awhile, sees what “driving” is, cataloges the characteristics of driving, and applies these characteristics to the present situation it’s in. Therefore, it doesn’t have to know everything, it just has to know what to do if something similar, or parallel happens.
Stanley doesn’t “learn” how to drive perse. It uses what’s known as “probablalistic robotics” to analyze the error in the data comming in from all sources, and then uses the most accurate data. I don’t know much more about it, but Stanley did push the speed limit a number of times during the race.
I really enjoyed Servo’s (or was it Nuts and Volts) articles on a FIRST team last year.
My only problem was in the calendar of events it made it sound like we build robots out of kits that are of the follow the directions to build the robot variety.
