Slashdot Mirror
DARPA Challenge Prize Money Restored
antispam_ben wrote to mention that, some three months later, DARPA has been able to find the money to offer cash prizes once again. The DARPA Urban Challenge will go forward next November with more than $3 Million on the line. From the article: "The race will see as many as 90 teams 'drive' an unmanned robotic road vehicle through city traffic, competing to finish a 60-mile course within six hours. Set for November 3 of next year, the challenge will call on robots to safely obey traffic laws, negotiate busy intersections, merge into moving traffic, avoid obstacles and navigate traffic circles. DARPA has yet to disclose the race location, but has said it will be in the western United States. The government research group didn't unveil the 2005 Grand Challenge location in the Mojave Desert until weeks before that race, in order to avoid giving any team an advantage."
Just hit the cruise control, and go to sleep! It's a less expensive, and a whole lot more fun!
I wonder who's going to be driving the other cars? In the previous races, the robots were traveling through a closed course with no traffic.
I sure hope it's a closed course, because I'd hate to be t-boned by an errant robotic Touraeg.
Isn't that dangerous? Is that even legal? Don't cars have to be certified for road safety before they can travel on city roads?
It should be. I bet it will be. That's not what the article says though.
10 miles an hour? Sounds like L.A., California.
DARPA San Andreas baby!
Actually, what I meant to say is that I'll be playing San Andreas in the automated vehicle while it safely navigates traffic--something I can no longer do after playing the GTA series. It is just too tempting to run down pedestrians and try to steal nicer and faster cars!
Will additional points be awarded if they successfully navigate the LA aqueducts, find Sarah Conner?
navigate traffic circles.
No American is going to win this one...
If I have been able to see further than others, it is because I bought a pair of binoculars.
But after much complaint from contestants, Kenneth Krieg, undersecretary of defense for acquisition, technology and logistics, approved the prize money.
No doubt the driving force behind this decision came from the folks at DARPA. First congress tells them to develop autonomous vehicles, then it proceeds to trip up their efforts with the "John Warner National Defense Authorization Act".
What I'd really like to know is why they're pushing this technology so hard and fast. Does it make sense to go straight to an urban environment when only four constestants even managed to finish the last challenge?
I always mod up spelling trolls.
but, while as a software engineer and electromechanical hobbyist I fully appreciate all the challenges involved with these robotic drivers, I'm just not impressed by systems that have courses plotted into them and use GPS and high resolution maps and intimate fore-knowledge of the landscape, etc. As a driver, -I- don't need that fore-knowledge to get from Sacramento to Manhattan - thousands of miles successfully navigated without any more fore-knowledge than that I have to travel generally North East through many states.
I will be impressed when driving automation systems can start with a general idea of where their source and destination locations are and can read the signs to figure out how to get there. They must use perceptive powers to avoid colliding with other drivers or running down pedestrians and following the rules of the road instead of range finders and lasers and GPS-based speed limit adherance and other such nonsense.
Until the system can be boiled down to a pair of eyes and a pwerful set of smarts driving , in my view, it's just an elaborate obstacle course being followed by the likes of this robot. I understand "baby steps", but "they" tend to avoid tackling these big challenges and instead continue to focus on these contraptions that just, plain aren't smart enough.
IMHO, of course.
This cannot be allowed to happen! Tell those DARPA spooks to take their ROTM challenge elsewhere! Or at least just flat out say it'll take place in LA and if mad robots happen to knock down every building in town, so much the better. I for one will avoid any city on that date.
political_news.c: warning: comparison is always true due to limited range of data type
But after much complaint from contestants, Kenneth Krieg, undersecretary of defense for acquisition, technology and logistics, approved the prize money.
Policy that is so prone to failure is about as ridiculous as a system that cuts off funding to an entire branch of the military if someone tweaks some minor policy somewhere.
These prize awards aren't just some minor toy program -- they are the future of technology development which means defense preparedness. Maybe there are some radical Muslim cleric moles posing as policy makers. Oh well... Islam isn't as bad as some theocracies.
Seastead this.
Ooo man the floppy drive is broken. No wait. The computer is just upside down.
Man I hope they televise this event, I dunno how many entries they will allow but it would be a real life version of Carmageddon (the smashing into stuff part, hopefully not the blood and guts) I would pay for PPV =) (if I had cable or dish)
-You have been modded appropriately-
Does the robot vehicle hit and run or stay?
Does it recognize a human laying in the road?
She was like chocolate when she drank... semi-sweet at first and then increasingly bitter.
If you'd seen the talk at google by last year's winner, you'd know that it's not as simple as you've made it. They're given GPS coordinates only shortly before the contest begins, in an effort to make sure nobody tries to precalculate the entire path etc (not sure what the input will be this time around). Not that it matters, as it would be hard to determine navigability with a simple 2d overhead satellite image. And they still need collision prediction & avoidance to avoid pedestrians and traffic (not that I'd be planning to drive in that town the day of the event). As far as speed limits based on GPS, that won't work out so well; regular GPS has an error of something like ten meters; that enough to confuse a highway with it's frontage road. But speed limits themselves raise an interesting question, if an autonomous system calculates that it can safely travel higher than the speed limit under current conditions, traffic, quantum states of nearby objects etc, what does that say about the limit?
Really, I somewhat suspect that recognizing roadsigns wouldn't be terribly hard. They're always placed with high visibility as an intention, and many are common among the states. But given the existence of road maps and map nagivgation databases, the only reason to include the feature is as a double check on your data; sometimes it's just wrong, but sometimes construction fucks up the map systems. I'd rather have both, frankly. You're definitely right that path finding over a map is a fairly solved problem in AI.
But most of all, I'd be fairly impressed if you made it to Manhattan without any knowledge of maps whatsoever, human or not. I live in Kansas, about halfway between those two points and neither Manhattan or Sacramento are on any roadsigns here.
I Browse at +4 Flamebait
Open Source Sysadmin
<begin obscure OpenBSD reference here>
They probably took the money away because they found out that Theo was Canadian and might enter the contest.
They restored the money when Theo confirmed that he wasn't entered and the money couldn't be used anyway because of conflicting license terms.
<end obscure reference>
For other brits out there, I've just googled "traffic circle" and I can confirm that the yanks haven't just made up another term for "roundabout".
No, in this case they've also entirely buggered up the fundamental design too!
http://www.alaskaroundabouts.com/mythfact1.html
Ooo man the floppy drive is broken. No wait. The computer is just upside down.
The best robot made something like 5 km. In the second race 3 robots (if I remember correctly) were able to finish the race. I wonder if history will repeat itself. First time a utter failure and second time some make it. There's a great documentary about this race in a torrent site near you..
The race will see as many as 90 teams 'drive' an unmanned robotic road vehicle
I know... I know... they did put 'drive' in semi-quotes, but it's still misleading to a reader who is unfamiliar with the Challenge.
I've said it before and I'll say it again: The robots will be driving themselves.
This type of design is worlds different from a system to be 'driven' using a joystick or by some guy monitoring the robot's progress. Amazing leaps and bounds in artificial intelligence, software image recognition, spatial on-the-go mapping, etc. are coming out of DARPA Urban Challenge that would never be necessary if there was a human hand--even remotely--behind the controls.
>> Standing on head makes smile of frown, but rest of face also upside down.
Are you asking about the right thing? First, you'd be surprised how much foreknowledge you probably have about driving from Sacramento to Manhattan. I've never done the drive, but I'm pretty sure if I hop onto a big, even-numbered interstate, like 80 or 90, I'll get most of the way there. And, if I find myself somewhere in the middle, I might not know the roads, but I'll have a decent sense of where each city lies. So, maybe it's "cheating" to give the robot a gps and a street map of everything, but the difference is one of degree. This background knowledge is "a pwerful [sic] set of smarts." Second, laser-range finders and gps-based spedometers don't give the robot capabilities you don't have--they let it do what you can do. You have binocular vision, and your brain works really hard to estimate ranges from that. You probably know how fast you're going because you use your eyes to look at the spedometer. So, these tools give the robot teh equivalent of your "pair of eyes." Not saying that a robt that can do this is the robot that will replace humanity, but I think this meets your test.
If I were a contestant, I would make sure my vehicle stays out of the way of Oshkosh Truck's entry! http://www.darpa.mil/grandchallenge/Teams/Track_A_ Teams/TeamOshkoshTruck.asp
I could see them doing unmanned vehicles in the desert, but city traffic? I don't think we are ready to travel on unused city streets yet let alone with other cars. Wherever they're going to hold this, I will be avoiding. I have enough tension following a car with that "student driver" placard on the roof to be dealing with a blender driving in the lane to my left.
I realize part of this challenge is to force the evolution of the technology and give it a push, but there are limits, and I think this next one exceeds them in a dangerous way. "learn to walk before you try to run".
I work for the Department of Redundancy Department.
good link! Mod up!
..until it can demonstrate the ability to shave and/or apply mascara at speed, cut off other drivers, and give the finger to old men going 48 mph in the passing lane. There's a good reason they're not holding this on the East coast.
Mission: To provide products that consume time and energy as entertainingly as permitted by the laws of thermodynamics.
Openness about everything is actually much better for avoiding any advantage to any team! Keeping it secret just opens up the possibility that one team will get an advantage through a backdoor channel. Openness = fairness.
Of course if you win this race you will spend the rest of your life doing things in secret in our secret underground laboratories.
David W. Hogg -- assoc prof, NYU Physics
I think the troops in Iraq have demonstrated the way to win
this competition, as they navigate urban streets
every day.
1) Always keep the accelerator floored
2) Mount big weapons on your vehicle, and use them occasionally
on threatening vehicles.
3) Use deafening loudspeakers, telling people to get out of the way.
4) Be willing to lose a couple.
Small problem of having the landscape rise up against you.
Although in the end game, the scenario would probably be that
locals might not want the vehicles to arrive at their destination.
So your robot vehicles have the same problem as human vehicles
today in Iraq, and I would think would need to adopt
similar strategies?
Why develop a car that acts like a good citizen in
commuter traffic, when this is about war vehicles?
but how will Robots negotiate Women Drivers???
Eclipse PDE and Me
That is, lanes that are designed to be negotiated by robot-cars. In the UK we have cats-eyes already that miark lanes for normal drivers to follow. I dont see it being much of a stretch to have digital markers, that not only show "visually" to the onboard computer where they're supposed to be "now" but can also, like road signs to humans, give clues that "in 100m there is a Stop sign" for example, maybe all markers could be placed 100m before they actually happen as a standard. so to summarise, instead of trying to teach robot-cars to follow human road-navigation cues install, in parallel robot-road-navigation-cues that are standard. (though one can imagine the standardisation would be the longest, hardest part, by which time the tech will be rock solid)
If you don't risk failure you don't risk success.
Ok... but this may seem silly to some, but after living in New York for the first half of my life, I may have a different perception of what city traffic is. New York City (or Manhattan Island for the purpose of this comment) Is about 8 miles from Battery Park to the Bronx (by my rough google earth measurements). This is about 13% of the distance which would need to be driven by these cars in the total course of 60 miles through city traffic. I don't believe I've met a person (including the local cab drivers that look like their pumped up on speed) that can in fact drive these 8 miles through New York City during traffic in less than 1.5 hours. Meaning that they would have to make an 11.5 hour drive in 6 hours that a living breathing driver couldn't do.
This is obviously why the money has been put back up, without actually bending the space time continuum, the prize can never be won.
I graduated from Tech back in 2004. At the time, the Robojackets student org wasn't up to entering DARPA. Nice to see an organization on campus ready and trying.
Good to see some names I know on the list of people, too!
Good luck!
This
I'm even more interested in the just announced DARPA project entitled "DARPA 2006 Grand Theft Auto".
The plan is to set up a course in downtown Detriot or Cleveland and challenge entrants in two categories: the total number of convenience store robberies and the average take per robbery.
They have yet to decide if the automated cars can use weapons of force in this competition.