I don't know how many of you follow the Darpa Grand Challenge news, but I watch it like a hawk, because I find the whole autonomous vehicle technology extremely interesting. They had a press release today naming the site and the ones who qualified to compete for the Darpa Grand Urban Challenge. The site is going to be the now-closed George Air Force Base, in Victorville, California and is going to take place on November 3, 2007. Here (.pdf file) is the official announcement with the entire list of qualifying teams, place, etc.. (37 semifinalists, of which 20 will compete in the eventual competition after the next cut)

For those of you that aren't familiar with it, it is a monetary competition ($2 million for first place, $1 million for second, $500k for third) in which teams will attempt to have their vehicles maneuver autonomously (i.e. without a human driver) through a setting that is typical in an urban environment. The vehicle has to be aware of other vehicles, road signs, etc. The last competition was the regular 2005 Darpa Grand Challenge where the vehicles had to maneuver through a desert road for a 212.4 km (132-mile) long course, which was pretty astonishing because the previous year (2004 - the first year of the competition) the furthest team made it 11.78 km (7.36 miles). The technology in this arena is really growing in leaps and bounds.

For more information on it, I would recommend going to the wikipedia page on it.



This post has been edited by Live Forever: Aug 10 2007, 06:23 AM

For anyone that hasn't seen it, I would recommend also watching the Google TechTalks presentation by the leader of winning team (Stanford, and the car's name was Stanley, haha) from the previous Darpa Grand Challenge (the 2005 one), Stanford computer scientist Sebastian Thrun, which is one of the most interesting things I have ever watched (it was filmed in late 2006 when they had just started working on the Urban Challenge vehicle):


http://video.google.com/videoplay?docid=85...517128412883394

He makes reference to the NOVA show about the race, which can be found here. (and watched online)

There are a lot of great benefits of this type of technology to society, more than what you would think of that he discusses, along with the tech involved in accomplishing this type of feat, the different approaches taken by different teams (I am glad Stanley won, because it seemed like the most sound strategy) and what it felt like to win.

This post has been edited by Live Forever: Aug 14 2007, 07:25 PM

By the way, in this latest one, the teams are going to have to adhere to all California traffic laws, merge into traffic, avoid moving traffic and other obstacles, negotiate intersections, etc. etc. to complete a series of "missions" in order to win. It will be much harder than the previous competition, but it is really exciting that cars are this close to being able to drive autonomously. Of course, Darpa wants it for military applications, but it won't be long, I suspect, before we start seeing it in production cars we can buy.

Yes, DARPA's Grand Challenge and Urban Challenge are indeed exciting. One of the most exciting aspects of the challenge, though, are the successes of the model for promoting scientific research. I'm not as libertarian as I once was and realize that there are many aspects of scientific research external to today's marketplace.

I think that, as autopilot technologies increase in complexity, we can aid some of the inherent difficulty of the unpredictability of other cars by legislating required beacons and central server communication. While it creates a greater networking problem, the AI problem of driving all cars is much simpler than the AI problem of driving one car in a pack of clumsy humans. To think, as we quickly approach a period of energy independence, we might exchange one chain for another, informational codependency.

How soon do people think it will be before stuff like this is actually implemented? I don't know if someone will 'pass' the Urban Challenge this year, but likely they will the next time they do it. And then how long before costs come down enough to implement it across the country? The only thing that'd cost a lot of money in the system (besides possibly proprietary software - boo!) is the sensor arrays. Fortunately, given the interfacing revolution, metaversal cartography, etc, prices will continue to drop.

My only fear is that this technology will arrive before sufficient battery/solar technology for electric cars and that people will commute even more unnecessarily.

By the way, quantum computing technology is starting to pick up, too. Just imagine the possible utilitarianism of a higher ordered traveling salesman problem with all of the cars, road capacities, and destinations all calculated at once, in or approaching real time. Geekgasm!

The thing is, it would save a lot (on interstates especially) as far as gas mileage is concerned. The computer would drive much more efficiently than a human driver, as well as having less stop and go naturally because the cars would communicate with each other and stuff, there would be more efficient use of space on the interstates (he goes into this a bit on the above video) because more space would be taken up on the interstate by cars because of the improved efficiency (which would also more than double the capacity for our existing roads without having to build new ones which would cut down on traffic jams), as well as the drafting effect (where you get better gas mileage from going behind another car or a big truck) because of the air being moved out of the way which given a computer driver would be less dangerous than if a human did it, and probably a half dozen other things I am forgetting at the moment that would improve gas mileage. Of course, solar tech would be great as well, but it would cut down on gas mileage significantly.

Of course that is only the secondary effect to the greatly improved efficiency gains we would have as a society if we could get things done on the way to work in the morning and evening, and didn't have to worry about driving, if we didn't have to worry about parking any more (just get out of your car and it goes and finds a parking spot and then comes and gets you when you are done), if we didn't have to drive our kids to school or soccer practice (the car did it for us), if we could sleep during long trips across the country and stuff, if we could cut the accident rate of motor vehicles down exponentially so that it would virtually be 100% safe, if we could.... (I am sure you get the idea and can think of a bunch more applications that would improve lives)

When this tech gets mature, it will have so many improvements for society, it isn't even fathomable at this point, and unlike a lot of the tech discussed in these forums, this is coming up really quickly. (there will probably be a car either this year or next that passes the Urban Challenge, which would be monumental in implications about how far the tech has come in only 4-5 years when it was virtually non-existent)

I see these types of competitions as being very libertarian. (at least economically libertarian) Offer up a prize and let the free market work things out. (of course, the market being more about ideas here, but still the same concept) This is a far more efficient and less costly way of developing the technology than spending the money directly on the technology. The prize for the first Darpa Challenge was $1 million for first, and the total prizes for this Urban Challenge one are $3.5 million (2 + 1 + 0.5 as stated earlier), and all of the progress would have happened in 5 years or less. (probably 4 years if they do it this year) If they had paid a government contractor the same thing to build a system like that from the ground up for military purposes (what their goal is), it would have probably been a few hundred million a year for 10 or 15 years, and the result would have been much less efficient than what you are going to get out of all these teams competing. Using competition to accomplish goals is one of the core tenets of (economic) libertarianism. Other branches of the government (and corporations for that matter) should take notice. If we have to have a government for some things (military defense being one) then I want them to use these types of free market principles as much as possible.

I'm thinking of going up for the final on Nov. 3. If someone wins the first time around it could be a while until something like this comes along again. If Stanley used 7 laptops to win with nothing but desert to interact with, California traffic will probably require several hundred laptops... or not. Guess we'll see.

The Nova program video I mentioned above (that was referred to in the Google TechTalk by Sebastian Thrun) has been recently added to Google Vids, so I thought I would embed it here for anyone interested (note: it aired before the TechTalk):



http://video.google.com/videoplay?docid=87...876587754396524

Here is the approach MIT is taking to designing a car for the Urban Challenge:

I doubt anyone will win this year, but I would love to be proved wrong.

Competition is in 5 days!
(On November 3, 2007)

It appears that only 11 teams will be competing in the final event. (all the rest have been eliminated) There were 20 slots, so this is a bit surprising, but not completely unexpected.

The 11 teams (out of the 36 that made the semifinals) are the following:
1. AnnieWay
2. Ben Franklin
3. CarOLO
4. Cornell
5. Intelligent Vehicle Systems
6. MIT
7. Stanford Racing
8. Tartan Racing
9. Team Oshkosh
10. Team UCF
11. Victor Tango

http://en.wikipedia.org/wiki/DARPA_Grand_Challenge
http://www.darpa.mil/grandchallenge/
http://www.tgdaily.com/content/view/34665/113/

Ohh it feels like a sporting event

I'm heading up tomorrow! Hopefully I'll be able to get some good footage to share although I don't yet have a firewire cable :(

I'll be looking forward to it! I bet it will be a fun time to be there.

Evidently people can follow the live video feed via the webcast on the Darpa Urban Challenge site on race day.

On their site:

...and via the web document:


So, it will either start 6:30 or 7:30 AM PT (thats 9:30 or 10:30 AM ET) and go throughout the day, I suppose.

By the way, in case people didn't know, the two main sites that have been giving coverage of the event over the past few days (weeks?) are:
http://www.tgdaily.com/content/view/34594/123/
and
http://blog.wired.com/defense/urban_challe...enge/index.html

Lots of recent videos and stuff if you click around on those sites. (I especially liked the videos of Stanford's team describing the tech they used) I am not certain if they will be doing live updates throughout the day or not on Saturday, but I'll be tuning in to the live webcast (mentioned in the post above this one) to watch everything unfold.

It is starting here in a few minutes it looks like. The webcast video has a countdown that is about down to 12 minutes or so right now before coverage starts.

Go to http://www.grandchallenge.org/darpauc07/watchtherace/
To see the webcast, status board, or tracking map, throughout the day while it is updated live.

Video is going now. Looks like there is some kind of a problem with the high bandwidth version, but the low bandwidth version is going good.

Haha, they have Jamie from Mythbusters as one of the sideline announcers. If anyone wants to watch the low bandwidth version (the high bandwidth one is still being finicky, at least for me), then you can open Windows Media player, and then click File-->Open URL and then enter this into the bar:
http://mantech-duc-low.wm.llnwd.net/mantech_duc-low

and hit Enter.

Go ad free, join ImmInst as a Full Member.