DARPA Grand Challenge Updates

Red Team writes "Today is the day. The official race route for the DARPA Grand Challenge was released to the first five teams at 4:00AM PST this morning. Our race planners are pouring over the race route getting ready for the launch. H1ghlander will start first at sunrise, around 6:15AM PST, followed by Stanford and then Sandstorm. For real-time updates on the race, you can track the Red Team race-day blog or catch the webcast on the official Grand Challenge page." Update: 10/08 20:57 GMT by Z : USSJoin writes "Stanford Racing, home of Stanley, has just finished the 131.2 mile DARPA Grand Challenge course. Considering that the CalTech Vehicle (Alice) jumped off the track toward onlookers only 8.3 miles in, this demolition derby-meets-AI demo has certainly been exciting."

Comments

Popular Science has most recent updates

[technoextreme]

http://www.popsci.com/popsci/darpachallenge/1b7a1e 7eef0d6010vgnvcm1000004eecbccdrcrd.html Apparently, the teams all ready have traveled four times as far than last year. It is great progress for a year.

Re:Popular Science has most recent updates

[Capt'n Hector]

Is this because they've been perfecting technology (no doubt they have) or because DARPA has chosen an easier route?

Re:Popular Science has most recent updates

[macklin01]

From

It's one of two entries by Carnegie Mellon. The other, a modified red Humvee dubbed Sandstorm, took third position in the trials. It was the best performer in last year's race despite covering only 7 1/2 miles of the 150-mile course. The exact route of Saturday's race will be kept secret until two hours before start time, but organizers have said it will begin and end in Primm and is expected to be more difficult than last year while covering as many as 175 miles.

So, it's a bit shorter (by 18 miles), but expected to be more difficult. -- Paul

Re:Popular Science has most recent updates

[Anonymous Coward]

Every team has managed to make it past the point traveled last year - except for one. The MITRE Meterorites, who in one hours and seven minutes traveled a total of 0 miles. I'll bet you they can at least double that next year. :) I have to wonder how they managed to qualify for this round if they weren't even able to leave the starting line.

(MITRE is probably best known as "that contractor who decided that Windows NT should run that battleship that was stranded when Windows NT BSODed on its test run".)

Re:Popular Science has most recent updates

[mroch]

Found it, on the Grand Challenge home page: "Kurjanowicz also said that the course 'is tough, tougher than last year,' but added: 'The vehicles are better, smarter. That's why we held the NQE [National Qualification Event],' he said. 'We wanted to make sure they are capable of completing the course."

Re:Popular Science has most recent updates

[Jay Carlson]

(MITRE is probably best known as "that contractor who decided that Windows NT should run that battleship that was stranded when Windows NT BSODed on its test run".)

Cite? MITRE didn't exist when the final US battleship was built, nor did MIT Lincoln Labs. I suppose the MIT Radiation Laboratory was contemporaneous in 1944, but I expect their expertise in OS recommendations was limited. I suppose this lack of knowledge of operating systems is excusable as there weren't any operating systems.

Oh, you mean the USS Yorktown? That's a guided missile cruiser, and back in the old days the hull would have been called a destroyer, before the Navy decided to change the nomenclature. Little bit of a difference between a destroyer hull and a battleship, but hey, AC abuse is par for the course.

As much I relish the image of some poor ensign yelling, "Screen's blue, SIR!", nobody seems to think this was an OS-level crash. And most of the google hits I can find on "navy smart ship mitre" point to things like Think Outside The COTS . Scrolling down to Figure 1, there's a list of potential pitfalls of commercial-off-the-shelf software.

If this seems familiar, you've been a slashdot reader for a few years: MITRE Corp. Report On Open Source In Government .

Re:Popular Science has most recent updates

[some2]

I was there.... just got back home. Most said the course was as challenging but less difficult to pass. On this course, the first stretch was a run over a dried lake bed to Jean, NV from Primm. Basically, if your vehicle was road worthy you should pass back to the start/finish point for a quick photo op. That was around the first 8 miles of the course. I saw one bot (think it was terramax) get locked up trying to go up the railroad overpass (B), but most bots handled that well. Most everything else was fairly out of view, so it's hard to say. The hard part was the Beer Bottle Pass (K). A few teams were complaining that their heavily ruggedized vehicles were not being fairly treated because the course was made so easily passable. More agile vehicles had an advantage in that they could get up to speed faster. There were no large boulders, trees, etc in the way of the bots, as they were supposed to stay primarily on the predefined trails. The GPS tracking had around 1000 waypoints according to several different members of a few of the teams. The Stanford bot "won" in a sense as they were the first across the line. The red bull ice containers were poured on their coach and the media was interviewing their team. The CMU/Red Team groups appeared pretty sad after the initial results came in, but the Red Team Too bot may have beaten Stanford.. we'll see. The way they are scoring this is painful because they constantly paused the bots in movement to allow for sufficient space between. The exception may have been when stanford's bot passed CMU's 'highlander' bot.

No webcast

[mpn14tech]

Unfortunately there is not a publicly accessible webcast of the race like last year. I was quite disappointed at that.
Still it is amazing how well the race is going this year. I hope there will be more races with greater challenges.

Re:No webcast

[pallmall1]

Even without a webcast, it's better than this year's OU vs Texas game.

We need a Google Maps Hacker

[Maddog Batty]

Can someone combine Google Maps with the XML here: http://www.grandchallenge.org/data/location.xml

Re:We need a Google Maps Hacker

[aerodyno]

Here's an Excel spreadsheet that displays that data and automatically updates every minute (it's a start)

http://www.cs.princeton.edu/~savraj/gc-live.xls

Re:We need a Google Maps Hacker

[vespazzari]

Here are the close up sattalite pictures from google maps of all the waypoints on the route information section of the darpa site.
Grand Challenge
2004 DEG Milestone, waypoint a
Railroad overpass, waypoint b
Lucy Gray Mountains, waypoint c
Roach Lake, waypoint d
Sheep Mountain, waypoint e
Jean Lake, waypoint f
Railroad underpass, waypoint g
northern tunnel, waypoint h
Southern tunnel, waypoint i
jean, waypoint j
Beer Bottle Pass, waypoint k

Re:We need a Google Maps Hacker

[pooya]

Kinda late, but here it is. Map is working: http://www.pooyak.com/utils/dgc2005/

Comment removed

[account_deleted]

Comment removed based on user account deletion

First races...then the WORLD!!

[Druox]

Sure, they're all just happy cars driving a fun race...until they rule us all!!
I, for one, welcome our four-wheel and rear-wheeled drive overlords, and pledge my allegiance to Emperor Camry.

Re:Sensors sensors sensors

[mroch]

Did you read anything about this? The course is specifically designed to prevent teams from relying solely on GPS by making them go through tunnels where they will lose their signals. Many of the bots have cameras, lasers, radar, etc.

Re:Sensors sensors sensors

[Maddog Batty]

They are a lot more clever than you think they are. The qualification event was all about avoiding obstacles that were in the way and had sections with no GPS coverage.

A couple of teams are using stereo cameras to find there way around but the vast majority are using SICK scanning laser range finders. They show you where things are over short distances which enables you to avoid obstacles at low speeds. Red Team are also using radar to be able to detect obstacles at greater distances to enable higher speeds. Unfortunately, it doesn't pick up everything. However data fusing everything together (GPS, INS, Radar, Lidar + others) allows you to move at modist speeds using current tech. Luck is always useful but technology is what is being used today.

TGDaily.com also has a blog up

[not5150]

http://www.tgdaily.com/2005/10/08/darpagrandchalle nge2005/
With pictures :)

Most interesting one so far is when Caltech's Alice charged through a k-rail, knocking it over and then started up a berm towards reporters. It was E-Stopped just a few feet away from hitting the media.

Re:TGDaily.com also has a blog up

[ikkonoishi]

Good thing they stopped it before it tasted human blood. If it had we would have never stopped its killing rampage.

Luck has nothing to do with it !

[timeToy]

Welcome to the future ! The time now is 11h45am PST, I can predict that several cars are going to finish the course this year, one might have been luck, 3 or 4 for is not. Sensors, but more than that, the actual programs that run the car is the key here. Heck even my own personal Lego Mindstrom creations can navigate their way inside my home ! mmm I may enlist a Lego based car bot next challenge !

any of the contestants here?

[Edmund Blackadder]

During last years challenge all contestants were complaining that the thing was rigged in order to get the red team to win. I.e. there were last minute rule changes for which the red team was prepared for, but nobody else was, and some contestants said they had an operational system ready yet they were not allowed to compete.

I am not at all surprised of this, since the red team is sponsored by the major military contractors and we all know how they basicaly control military procurement.

But I was wondering if similar shenanigans were happening this time around. Any of the competitors care to comment?

Re:any of the contestants here?

[mroch]

Maybe it has nothing to do with favoring Red Team, and more to do with the fact that CMU has one of the top robotics and AI university programs in the world, teamed with defense contractors that should be expecting something like that, so they just happened to be more prepared...

Re:any of the contestants here?

[Edmund Blackadder]

"That does not make sense. We are talking about a cash prize of 1 million dollars (last years). What does winning one mil do for a defense contractor?"

nobody really cares about the 1 million ... they care about the contracts that will start flowing after a succesful vehicle has been made, and the army orders 50 000 autonomous trucks. It is obvious that the winner will be the one that is favoured to get the contract.

Re:any of the contestants here?

[Animats]

As a team leader of one of the teams eliminated at the NQE, I didn't see any visible favoritism by the DARPA staff. The teams that went to Primm are the teams that should have gone.

Funding is more of an issue. Teams were supposed to have no Government funding whatsoever, either direct or indirect. Yet MITRE had a team, and they're a quasi-governmental agency. CMU has received DARPA robotics contracts for years, as has Stanford. Red Whittaker of the CMU team is still the principal investigator on a NASA grant (#NAG5-12890) until February 2006. Stanford used software developed under DoD contract, although anyone can download it and they asked DARPA for permission. It's more of a revolving-door issue than direct diversion of Government funds.

But the real incentive for the big university teams was fear. If Joe's Auto Parts fielded a better robot than some university getting $20 million a year in robotics funding from DARPA, DARPA might well pull the plug on the school. CMU faced that prospect; originally, they weren't going to enter the Grand Challenge at all. The whole Grand Challenge was created because of unhappiness at DARPA with the rate of progress in mobile robotics. DARPA has been pouring robotics money into CMU and Stanford for thirty years, without getting much back. The head of DARPA, Dr. Tony Tether, decided that it was time to do something about that. It worked.

Re:Live video feeds?

[mroch]

There is a webcast on campus here (Carnegie Mellon):

"A live Webcast of the Oct. 8 Grand Challenge through the Mojave Desert will be shown on campus in Breed Hall in Margaret Morrison Carnegie Hall.

"The Webcast will begin at 9 a.m. and end at 6:30 p.m. Breakfast will be served from 9 to 11 a.m. and lunch from 12:30 to 2:30 p.m."

http://www.cmu.edu/cmnews/extra/050927_redteam.htm l

Lovely image!

[HerrGoober]

"Our race planners are pouring over the race route"

DARPA's site for status update, not team sites!

[vectorian798]

Here:
DARPA Grand Challenge 2005

There is a map updated almost every minute automatically that will show you the position of all the teams and the times elapsed for each, etc. At the time of this post, Red Team Too, Stanford, and Read Team are all doing well, at 94+ miles each. Surprisingly, most of the teams are still in the running (that is, not eliminated). It is hard to compare one team to another however, because each team starts at a different time and perhaps the ones that are far behind are in the hardest part of the course and are thus moving slowly.

Re:DARPA's site for status update, not team sites!

[Maddog Batty]

There was a 10 mile error in the data from the start. Darpa corrected it which made it jump back as you noted. It is believed that the current data is more accurate read the thread here: https://dtsn.darpa.mil/grandc/forum/topic.asp?topi c_id=1652&forum_id=30&Topic_Title=Post+Time!&forum _title=Grand+Challenge+Event&M=False&S=

Soldiers: Yay!; Truckers: Boo!

[Saeger]

Last year the best car made it only 7 miles. This year all cars but one have made it at least 22 miles (so far), with three frontrunners past the 100mile mark (so far) and expected to finish.

Now that's some amazing progress.

This is great news for the soldiers soon to be removed the line of fire; "ominous" news for the millions of truckers and taxi drivers (in the US alone) who'll be quickly replaced over the next decade.

Wikipedia Article

[KrackHouse]

I'm updating the Wikipedia page. Updates with links to videos plus google maps - add stuff if you've got it.

http://en.wikipedia.org/wiki/2005_DARPA_Grand_Chal lenge

Down to three teams, approaching 1000 foot drop

[Animats]

Ten miles to go, three entries (Stanford and both CMU vehicles) are close together, with Stanford well in the lead. Now approaching the section with the one-lane dirt road with the 1000 foot drop on one side. Is there a live video feed of this area anywhere?

Stanford racing team has won...

[Harry Balls]

...with their entry "Stanley", a modified VW Tuareg.

Total time: 7 hours, 8 minutes for a distance of 132 miles, which amounts to an average of 18.5 mph.

Re:Stanford racing team has won...

[Torrenc]

Err, no. They just finished with a time of 7h 28m. Maybe you mistoke the 125 mile line on the status board on http://www.grandchallenge.org/ for the finish?

Re:Stanford racing team has won...

[deblau]

Red Team Too is the closest to the finish line, but they're over Stanley's time. Red Team (One) is at Stanley's time now, and they're up in Beer Bottle Pass, 7 miles from the finish. I don't know about adding penalty time, but it looks like Stanley has won.

Re:Stanford racing team has won...

[Xerotope]

It's not so much an issue of adding penalty time, but subtracting pause time.

Both Stanford and Sandstorm have been paused several times to prevent them from running into the back of H1ghlander. It seems the roads are too narrow in most places to allow passing to occur, hence they stop the rear robot to allow a safe following distance to accumulate. The time the robot is in pause state does not seem to be taken into account in the unofficial results on the grandchallenge.org site.

Stanford and the two CMU teams have finished

[Animats]

Stanford is about half an hour ahead on time, so they win. (The numbers on the DARPA tracking site are wrong; the clocks are still running for the teams that have finished.) Insight Racing, TerraMax, and the Grey team are still running, but they're not even half way and can't finish within ten hours.

So it's over.

Autonomous vehicles will never be a joke again.

Note On DARPA Times

[Anonymous Coward]

Just thought I would mention a bit about the "Live update" times on the DARPA Grand Challenge Site since everyone keeps referencing them and they haven't been worth a damn all day:

1 - The timers were started ~20 minutes before the bots took off for at least Red Team Too, Stanford & Red Team and never reset.
2 - The bots were sent out at 5 minute intervals in this order H1ghlander; Stanley; Sandstorm, but Sandstorms time as only been 2 minutes off Stanley's all day, hmmmm.
3 - As someone mentioned the official clock for each bot is stopped if it is ever paused by the chase truck but it is clear that since the clock for each of the current finishers is not stopped YET, that the "live update" times are not linked to this official timer.

So, unless someone is posting from Primm or DARPA has posted official finish time since I started this post we all should just sit tight till those times are out.

Although, since Stanley started second and finished first he is most likely the winner (no, I'm not from Stanford; actually from CMU).

Kudos to all those that competed and Congrats to those that finished!!

Re:Racing against the clock

[AxsDeny]

Their plans are surely similar. The guy heading up the Stanford team is a former student of Red Whittaker and was on the Red Team last year.