CMU First To Qualify For DARPA Grand Challenge

Anonymous Coward writes "As of 18:00 March 9th, Carnegie Mellon's Red Team is the only entry to successfully complete DARPA's Grand Challenge Qualification Inspection and Demonstration (QID) before the main event on March 13th. The NY Times has this article detailing this first step towards winning the Grand Challenge."

Comments

Mars Rovers

[BenBenBen]

Just out of curiosity, how well would the rovers' "route picking" routines cope with this challenge?

I read that the operator says "go from here to here" and the onboard 'AI' chooses the best route in a 3d visualisation - is this software open-source, and could it be used in this challenge? I can't see any major differences, other than the relative lack of parked cars on Mars (2 pathfinders and a beagle, iirc)

Should DARPA have emailed NASA before starting this? ;)

Re:Mars Rovers

[genneth]

Pretty well, except for the speed thing, and the distance covered. From what they say, an average speed of 25 miles per hour will be needed to even complete the course in time. The rovers can presumably plod along and if it gets stuck it stops and asks for human intervention. The rules of the competition designate that no communication is allowed. From a piece that I read somewhere like New Scientist, it seems that with 4 Itaniums and 4 Xeons they're still not computing obstacle avoidance fast enough. For the qualification the team ran the vehicle at a walking paced 5 miles an hour. There's gonna be a lot more work needed until they can manage to finish in time.

i'd be more impressed if

[theguywhosaid]

they divided your time by the cost of your machine.

its impressive when you build a mega$ robot, but a minimal robot that manages to finish is way cooler

Where the pictures at?

[Ethon]

TAIWWP :( Does anyone know of anywhere hosting pictures of these unmanned robot vehicles?

Re:Cost to PRIZE ratio.

[drspock]

The real payoff, as the Red Team and everybody else knows, is a future DoD contract, for many millions, or billions, of dollars.

How is this impressive?

[dave420]

I know it's much more complicated than this, but they're giving it the knowledge to navigate a route, not the intelligence to come up with its own route. Surely that's missing the whole point of this competition? I read in the last /. article that they're using a loophole in the rules to get so far.

Seeing as DARPA wants to turn this technology into a military robotic transport, I don't know how valuable it's going to be if it has to be pre-programmed with terabytes of data just to move. What about if they invade somewhere they don't have good maps of? Somewhere with a dynamic landscape (desert, rocks etc)?

I'm all for innovation, but exploiting poorly-worded rules just to win for winning's sake is an empty victory at best.

Red Team is the least impressive in some respects

[John Harrison]

Red Team is using "the best map in the world" to guide it. They have used topo maps, aerial photography, and a bunch of undergrads to painstakingly map out the terrain of the possible courses.

All competitors are given the actual route as a series of GPS waypoints a few hours prior to the race. Red Team is going to send those waypoints back to CMU, have the big iron there figure out the best course based on all the map data, and then download that course to the robot prior to the start. In a way this is cool, but it seems like they are using a loophole. A much more interesting problem would be to navigate a course that you know nothing about other than the waypoints.

The other teams are using techniques that require more onboard intelligence and route finding. The most interesting vehicle is from Cal. They have a motorcycle. Even though I went to Stanford I am rooting for the Cal motorcycle to do well since they have the most unique vehicle. Hopefully the team of Stanford alums (already dropped out) can come back next year and beat them.

picture comparison

[morcheeba]

Here's a size comparison from the Oshkosk website... their truck is 9 feet tall, a hummer is 6 feet.

Re:Cool

[dave420]

That's another thing - they only look the part - if a mountain popped up you'd have more chance getting over it by foot than a humvee. They use Tacoma or Silverado chassis and engines, with a much larger body shell (resulting in the insanely low mpg and silly look). Range Rovers are the complete opposite, however. Efficient and excellent offroaders. Why won't people learn? :-P

Re:Interesting, but ...

[Belisarivs]

But let's face it, this contest isn't about finding cheaper ways to haul cargo or reach remote locations.

Sure it is. Logistics are a *huge* problem for the military, especially one that moves as fast as America's. Remember in Gulf War II that some of the most public incidents of American losses involved supply convoys, not front-line forces.

With this sort of technology, supply-lines become more like conveyor belts than masses of convoys. They elminate the need to teams of humans to transport fuel, water, ammunition, etc. to the front lines. This increases the pool of human resources available to the military for other jobs, while eliminating the worry of casualities inflicted by enemy interdiction missions.

Sure, automatic tanks will logically be a followup, but I think the military's mid-term goal is automating the logistics.

Surprised a bit by this...

[vudufixit]

I'm somewhat surprised that so many young people would work on a project that will help our military develop unmanned hunter-killer vehicles. Isn't this why DARPA's funding this project? I'm not against the project, just curious if there are any conscience issues involved here.

AW&ST February 23rd...

[mykepredko]

You might be interested in knowing that, according to AW&ST, the army/air force in Iraq has found in many cases that it is more efficient to transport cargo within Iraq via C-5, instead of 12 large trucks.

The reason was because the loading and unloading areas could be secured but not the highways in between.

Check out the February 23rd Issue.

myke

Re:AW&ST February 23rd...

[mykepredko]

The article (sorry not online) just gives the example of one C-5 being able to replace "12 heavy trucks".

I would agree - I would expect a mix of C-17s, C-130s, CH-47s and CH-53s rather than using just C-5s.

Just doing a Google search on the C-5, it is rated (from http://www.fas.org/man/dod-101/sys/ac/c-5.htm) as having a "Fully Loaded" takeoff distance of 12,200 ft with a "Fully Loaded" landing distance of 4,900 ft. Maybe the strategy is to land a full aircraft and take off in a nearly empty one.

Even in this case, I would think a fully loaded C-5 (anywhere from 770,000 lbs to 840,000 lbs (same source)) would break up the pavement of any modest runway almost from the first landing.

Regardless, the point is that the military recognizes the dangers of sending loaded trucks along potentially unsecured routes and is looking for a way to keep drivers out of harm's way.

myke

Re:Cost to PRIZE ratio.

[coyotejoe76]

From articles I've read (unfortunately I can't recall where) the whole DARPA event was used to "scare" one of their existing research partnerships (CMU) into getting their @$$ in gear and producing an automated vehicle - which apparently CMU was lagging on. The event was to show that DARPA could get great research from other sources if CMU didn't shape up.

The result is that CMU stopped dragging their feet, which accomplishes the main goal of DARPA $1 million challenge.

Forged message

[Animats]

That message is not from Team Overbot. Nor is it a simple repost of things we have said previously, although it does include some of our content.

I will pay $100 for the name and address of the person responsible for that posting.

John Nagle
nagle@overbot.com