DARPA Grand Challenge Kicks Off March 13th
GillBates0 writes "A quick reminder that the DARPA Grand Challenge is due to kick off March 13, the coming Saturday." He points to this "quick recap of the teams participating in the event," as well as details about the available satellite feeds. "The Atlanta-Journal Constitution is running a story about the event today. Quoting Frank Dellaert, co-director of Georgia Tech's robotics lab from the article, 'I would have trouble driving some of these roads myself. I think it's beyond the capabilities of autonomous vehicles today.' (shameless school plug). We'll see if the participants can prove him wrong."
Iphtashu Fitz adds a link to the New York Times' coverage of the trans-Mojave race, whose participants include "among other things a seriously tricked out motorcycle. The race is being run by the Pentagon, who is offering a $1 million prize to the builders of the first robot to successfully navigate a 200 mile route across the desert. ... a blog on ScienceBlog about the race has just started as well."
I would have trouble driving some of these roads myself
The real problem is that his turn signal is on for 150 of the miles and confuses all the autonomous vehicles.
Red Team: "Muahaha! Bow before the greatest CS school in the universe and our giant Hummer that is smarter than you!"
Blue Team: "Feh! We can beat you with two wheels tied behind our backs!" *obscene gesture*
Red Team: "Come a little closer and say that."
How will these robots be routed around wilderness areas generated by the California Wilderness Protection Act?
Wasn't the Barstow to Vegas motorcycle race cancelled due to declaration of these same wilderness areas? How is DARPA ensuring these vehicles aren't going to run over some tortoise?
Dont' get me wrong, as I'm no tree-hugger. However, it seems the Wilderness protection act only applies to people who cannot afford a congressman...
At least two other teams have formally dropped out, and we expect some no-shows.
CMU is the favorite. Fifty people, $3.4 million spent to date, direct support from aerospace companies, and a team leader who expects people to work all night, day after day. (Read the article in the current Scientific American.) But their technology is rather disappointing. The whole route is preplanned by hand, using a bunch of people at workstations in a big trailer with maps obtained by overflying the route with LIDAR-equipped reconnaissance aircraft. It's not very autonomous. They found a loophole in the rules and exploited it very effectively. There's no breakthrough there.
Anthony Lewandosky, with his self-balancing motorcycle, has the most innovative technology. We've met him, and are impressed.
Palos Verdes High School has a viable entry, using a Honda Acura. We've loaned them some hardware. They've had autonomous driving working for months. They started by having handicapped driving control actuators put into a car, which simplified their mechanical problems. They debugged using a golf cart. Very nice work.
Caltech tried to qualify today, but their vehicle made an unexpected turn and bumped into something. They get a second chance on Wednesday.
Most likely, no one will finish. Nobody has really done enough field testing yet.
John Nagle
So how does the favored team train for a 210 mile race through the Mojave Desert?? Why, by testing it in the SNOW of course!
I don't think you're going to get six good laser rangefinders for $5000. I don't remember the numbers as well as I would like, but I think the current favorite rangefinders (I think the brand is SIC?) are well over $1000 each. And you will quickly exhaust your laptop's computational power just denoising the output from crappy sensors. Heck, maybe even for the best sensors.
Autonomous vehicles have already driven across the country on highways, 98.2% of the time without human intervention. The roads it drove on are (I'm guessing) likely to be much nicer than those in the desert. Furthermore there was a human available to handle the surprises. For humor value: I believe one of the self-driving vechicles from CMU has a learner's permit from the state of Pennsylvania. See No Hands Across America for more info on this project.
The hard part of any project like this is uncertainty in the environment. The road may "disappear" completly from your sensors, or you may spot multiple roads. Maybe some mica on a rock screws up your rangefinder. Maybe your vehcicle's transmission gets a little "funny" and you can't shift properly anymore (I saw such a comment attached to this article). And we aren't even talking about genuine malfunctions like a failing rangefinder or sticky throttle.
I think autonomous systems might be the best example of the best laid plans of mice and men not succeeding when the slightest thing goes wrong. In fact, Steinbeck's story seems directly analgous to the problems of self-driving vehicles.
-Paul Komarek