Slashdot Mirror
DARPA Grand Challenge Updates
GraffitiKnight writes "After only 1 team managed to successfully navigate the DARPA Qualifying course, DARPA has rewritten the rules to let almost everyone compete. Wired has the story, which also mentions rumors that the race will run to 150 miles, much less than the original plans of 210 - 300 miles." Here is some earlier Slashdot coverage of the race.
I hear that it will be the machines hunting a prisoner or something over a 150 mile course.
From the Previous Story
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.'
I guess he was right after all...
If I was the one team racing unless there was a time restriction I'd go Mars rover slow.
No sense taking risks when there are no competition.
...as the article makes them out to be.
All the Wired article states about the Caltech and Ohio State teams is that "The squads from Caltech and Ohio State University were also allowed in, even though their drones did not complete the obstacle course. "
From the Caltech team site: "Bob completed the test route flawlessly until the last few feet. He was stopped by DARPA officials seven feet away from the final obstacle -- although had he been allowed to continue, he may have stopped himself in time..."
Seems close enough to me.
But if DARPA pulled out because of substandard results, they run the risk of being bad-mouthed, and perhaps even losing cred.
And the next time around they conduct something like this, not too many people would be willing to compete.
However, this lets DARPA see more entries - agreed, some crappy ones - but a lot of good ones which are good but would have otherwise not made it. Besides, its really too early to say anything, so lets see.
DARPA put up a $1,000,000 cold hard cash for an autonomous vehicle and they got dozens of shoddy immitations that couldn't even navigate the test course.
Normally they have to pay a defense contractor BILLIONS to get something that doesn't work.
They saved loads of money, and they don't have to pay until it works, unlike, oh, other DoD projects, like the Osprey, Comanche, Patriot, TW Missile Defense, etc.
"Life's funny sometimes." "And sometimes it isn't." --Cat's Cradle
You don't accept substandard results if nobody can produce. This is something you intend to throw massive money on eventually, you'd want the would-be contractors to put up or shut up real quick.
Then again, this is very very early in development for all teams involved. Should they cancel the event or only have one team competing? Kinda ruins the whole purpose. DARPA set an ambitious goal and, seeing that the technology wasn't quite there yet, revised the goal. Nothing wrong with that. It encourages people to participate and, by allowing more teams to actually get involved with the competition, mistakes will be made, they will learn what not to do, and the science will advance.
Remember, they're not ordering a few billion dollars worth of equipment yet. This is mostly a proof-of-concept event to foster investment from outside parties. Start small, encourage teams to make advancements, then hold a more challenging event in a year or two. Seems like a good way to do it to me.
Want to improve your Karma? Instead of "Post Anonymously", try the "Post Humously" option.
I'm not so sure letting them compete is such a crime. It's not like they have agreed to a multimillion dollar contract with some company that can't produce. If the designs suck they won't win, no money out of DARPA's pocket.
However a poorly designed bot can have a desing feature that if developed by the right people would be usefull. This lets DARPA see how some of these potential inovations will perform.
As x approaches total apathy I couldn't care less.
-- You can't spell geek without a EE.
The results of the attempts of today's group break down as follows:
You mean like the 'Star Wars" Missile Defense System which has failed numerous tests of increasing ease, but is being used anyway because "it beats nothing", except that "nothing" doesn't violate treaties we signed, creates a false sense of security, doesn't motivate anyone to 'get it right' and wastes trillions of dollars.
Or the Patriot Defense System, which routinely targeted friendly aircraft during development, failed miserably the first time it was put into use(for a use it was never intended- it's never been used for what it was originally designed for, shooting down planes) and then 10+ years later was used again and resulted in the deaths of dozens of UK soldiers because it couldn't tell the difference between a helicopter traveling at less than 100 kt and an enemy missile traveling over the speed of sound?
Or the Osprey tiltrotor, which suffered an astronomical failure rate and again, caused dozens of deaths of US marines?
Then there's the Comanche helicopter, which they've been kicking around for years and finally decided, after spending billions, to just say "oh well, so much for that"?
The defense department is famous for bidding scandals(if contracts are put out to bid at all), and being happy to look the other way and fudge the requirements(or ignore them completely) if the system fails to meet original requirements.
Curiously, the russians never quite had such problems. Their fighter jets, for example, don't require pristine runways and constant maintenance; they're built like tanks, because the people who designed them knew they'd be held responsible if it failed unreasonably...and responsible doesn't mean "loose their job", it means "end up in Siberia" or "in a river with a bullet through your brain".
This country needs three things. First, a true capitalist system for defense contractors. You want to sell the Army a tank? Fine. You can do so all on your own, without a single fucking dime, and then try and sell it. If it can't compete, too bad, your company goes under- that's the way capitalism works. Second, defense contractors need to be held responsible for when their products fail. Refunds for starters, contracts that can be invalidated on failure, civil/criminal punishments for gross design/construction failures. Third, absolutely, positively, no secret budgets of any kind. I am entirely pissed off with the pentagon filling up with all the kids who had secret treehouse clubs when they were kids and want to do the same shit now that they're 40.
Please help metamoderate.
The QID was pathetic. We spent two days watching vehicles move around at 1MPH and hit big, obvious obstacles. No way can most of those vehicles operate effectively offroad.
The big design mistakes seem to be these:
Only CMU is doing well. It's not the money, by the way. Their actual cash outlays are only about $300K to date. It's the body count and the fear. They have about fifty people on the project, a slavedriver boss, and the full backing of CMU. CMU has to do well; most of the Robotics Institute funding over the last three decades is from DARPA, and DARPA can turn that money off at any time.
John Nagle
Team Overbot