Slashdot Mirror
Stanford's Stanley wins DARPA Grand Challenge
tonyquan writes "DARPA has just announced that Stanford's "Stanley" autonomous ground vehicle has won the Grand Challenge, a $2 million contest for driverless vehicles over a 132 mile course in California's Mohave Desert. Stanley's winning time over the course was 6 hours, 53 minutes and 58 seconds, for an average speed of 19.1 mph. Second was Carnegie Mellon's Sandstorm (7:04:50), third went to another CMU vehicle "H1ghlander" (7:14:00) and fourth to the Gray Team's KAT-5 (7:30:16) More info from DARPA."
Looking at the final stats on the Grand Challenge website, it would seem that only five teams, out of the 23 that made the finals, were able to finish the course. The team that got the farthest before calling it quits managed about 80 miles, which means that the cut between those who made it and those who didn't was still pretty big. Another interesting thing about the final results is that, if you look at the pretty red and blue graph lines, they describe what looks like a sort of decaying function...
Or perhaps I'm just a dork.
No, but I can say Fahrvergnügen, especially since Stanford's team leader Sebastian Thrun is actually from Germany, you hit the nail on the head.
Great run, saw it on TV yesterday, and a major step in development of fully autonomous bots.
A World in a Grain of Sand / Heaven in a Wild Flower,
Infinity in the Palm of your Hand / And Eternity in an Hour.
Take a look at the route on the darpa web site.
Portions of the route DARPA set the speed limit as low as 5 mph. The highest speed limit on the course was 45 mph. The route included very very narrow passages, twisty bits along the side of mountains with 100 foot droppoffs and no guard rails, chunks with no roads at all going through gullies, tunnels with no GPS feeds where you have to navigate inertially and with sensors.
This was _hard_
The course did have a fair number of twists and turns in it. There were some places, like dried lake beds, where the cars could open up a bit, but for the most part it was bumpy dirt tracks one which even you or I couldn't do more than, say, 40 mph. There were also, intentionally, a fair number of obstacles designed to throw the computer systems off. You and I wouldn't have much difficulty in recognizing a cattle gate on a road, but imagine trying to teach a computer vision system to distinguish that. In other cases, the robots had to drive through tunnels that would not only be dark (making vision systems less accurate) but also lack any GPS signal.
So, yes, it did average out to a pretty slow "race." But, on the other hand, it is a marked improvement over last time, when no one even came close to finishing. I think that, in the interests of trying to ensure that they safely finished the course, let alone win, the various teams were playing it a little conservatively, and not trying to go for pedal-to-the-metal performance. Maybe next year, now that they have some confidence.
For far better info than the anemic (and completely flash based) gc.org site:
m l -- DARPA's GC message boardse nge2005/ -- Was updated throughout the actual event. Best coverage I've seen yet.
http://www.darpa.mil/grandchallenge/discussion.ht
http://www.tgdaily.com/2005/10/08/darpagrandchall
http://www.popsci.com/popsci/darpachallenge/ -- Popular Science's rather disorganized site
I'm still looking for "highlight" video myself... or pretty much any non-bland video (seeing them cross the finish line is nifty and all, but that was not a challenging part of the race). I particularly want video of Alice trying to take out some reporters!
Its just the first to cross. From the site: " .. this is a time elapsed challenge.. The winner will be decided after all the robots either complete the race or drop out "
http://www.rajeshgoli.com
One other thing to note is that this is hardly the first instance of cars driving themselves. Even in the late 1990's, there were several institutions that had programs to develop cars that could drive themselves. I believe CMU outfitted a minivan to drive coast to coast autonomously. There were some caveats, like having a human driver in the seat, hands poised over the steering wheel, ready to take control in an instance. I also think that, when the team was pulling off the highway to find a motel for the night, that was done by humans, too. But, for the highway stretches, where the car was operating autonomously, the vehicle was able to cruise at highway speeds (i.e., around 65 mph).
It's not so much an improvement in the AI as it is an improvement in the sensors. These vehicles look ahead about 30 feet and plot their course based on very simple logic. If there is a negative obstacle (a hole), it is more difficult for sensors to detect than if there is a rock sticking up in the path. Last race, the only thing that stopped red team was a hairpin turn. Their sensors looked straight ahead and only a little to the sides, but when faced with the hairpin turn, the vehicle almost fell off the side of the mountain! But the rules of the AI haven't changed much- just the sensors. If you're driving through jungle, for example, you have to have sensors that don't see leaves as obstacles. Otherwise the path will look totally impassable.
Funny building a car isn't that hard the power plant is rather complicated but the rest can be made of some nice tube stock and sheet metal for the most part. Granted it wont ride as nice as most commercial cars but it will stand up in an accident better than any of them. Granted I'm talking about good old fashion dune buggy with sheet metal attached. Never had any federal guide line issues just one state inspector made sure nothing would fall off and the wheels were covered.
No sir I dont like it.
sort of.
...) while i think palo alto has much better weather than pittsburgh :)
:) I would'nt be surprised if they also use large parts of the basic control and command software infrastructure (TCX) written by thrun and others while at cmu. if it is, no wonder they required
:)
the stanford leader (thrun) and their lead software developer
(mike montelermo (sp?)) were originally from cmu.
they only recently moved to stanford. although thrun claims it's coz of his wife, some people think it was coz of too much competition and bad blood at cmu which has lots of people working in mobile robots (wittaker, simmons, nourbaksh, choset,
the particle filter based localizer and mapper was developed while at CMU. Frank Dellaert (now at georgia tech) first introduced that to mobile robotics after reading about the
condensation algorithm in computer vision (i like to believe that i had a part in that last bit
7 PCs for redundancy, that is some of the worst spaghetti code i've ever had the displeasure of working with. it's easier to make it fault-tolerant by just throwing more hardware at it.
i'm not trying to belittle stanford in any way, but i just thought people might be interested in knowing that the real story in this case is a lot more complicated. the relationship between the winning teams were a lot more incestuous
thrun BTW is an amazing all-round guy with an infectious smile all the time.
Maybe in 20 years we can have auto driving cars that can make it so there is next to 0 car accidents.
Unless those are much requested "flying cars" there is next to 0 chance to create this for legal reasons. Families of walking city crowd killed by such cars would demand trillions from car makers each day. So, car companies will rather leave _you_ responsible. If auto driving cars are flying, thats another story. Without any way to switch to "manual" navigation, accidents could really be eliminated. Users would be allowed only to choose target location from pre-defined set on iPod-like dialer.
839*929
A friend of mine was on the CMU team. One of this years robots was in last years competition. Of course they made enhancements to the robot, but the biggest problem they had with it was that they rolled the robot 3 days before the competition. They were pushing the robot to the limits on a test track and went too far, according to my friend. He said they would have faired much better, but when a Hummer filled with computers rolls over, you are bound to have some problems.
by all means, don't let the facts get in your way either:
http://www.army.mil/fcs/articles/
http://www.uniteddefense.com/pr/pr_20050414b.htm
http://www.jointrobotics.com/history/MP89.pdf
So yeah, it seems utterly clear that the DOD has no plans to incorporate technologies for ground navigation into assault vehicles.
"Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
Actually, planetary rovers are just a tiny, tiny portion of the reason for this challenge, otherwise NASA would be sponsoring this, not DARPA. The primary reason for this challenge is for troop supply and support vehicles that can accompany troops into a battlefield, or be sent in autonomously. Which means the jungle scenario is non-trivial. One of the reasons the challenge is being held where it is, is due to the development lifetime projected force deployments being in mainly desert regions. Another major projected use for these kinds of vehicles is for deployment in a bio-hazardous area for testing and sampling in an autonomous measure. But once again, the is a DARPA challenge, not a CDC one.
"Unheard of means only it's undreamed of yet,
Impossible means not yet done." ~~ Julia Ecklar
Nice attempt to braoden and generalize, weasel. You said:
Clearly you're talking specifically about the Grand Challenge project, for which the time frame of "2010 to 2015" is relevant, and this is what I addressed. Stands to reason, as that's what thisIf a job's not worth doing, it's not worth doing right.
I do have to disagree with a comment by the parent: "...some people think it was coz of too much competition and bad blood..." I never perceived this and computer science at CMU is remarkably sparse in bad blood compared to other universities.
I go to Tulane and I can think of about 5 Tulane students I know who helped with the bot at one point or another including one of my good friends. Tulane was mostly helping out with programming and I know they were using Java. The story I heard is the sailboat Gray Insurance had been spending money on sunk, so they decided to enter the grand challenge for fun. Luckily the car wasn't destroyed by Hurricane Katrina (they were keeping it about an hour outside of New Orleans). Some of the Tulane students stopped working with the project after the hurricane because they are attending school this semester in other states (since Tulane is closed right now), but a few of the guys took the semester off to work on it full time. As a side note, I am impressed that anything my friend touched actually worked and did what it was supposed to since I personally know his track record of breaking things over the past three years (including MY CAR).