Slashdot Mirror
Carnegie Mellon Gets $14.4M to Build Robo-Tank
coondoggie passed us a NetworkWorld article, this one discussing new developments in the state of robotic warfare. Carnegie Melon is now hard at work on a tank set to join its brother, the already much-discussed Unmanned Areal Vehicle, on the modern battlefield "Ultimately unmanned ground vehicles would be outfitted with anti-tank or anti-aircraft missiles and anti-personnel weapons to make them lethal. Part of the new award budget is also slated to help the university prove that autonomous ground vehicles are feasible in future combat situations."
First, it's been obvious for a long time that robot tanks (and eventually robot infantry) are an inevitable development. It WILL happen, has to happen. Some of the posters will spout some meaningless garbage about how you can't trust a machine to decide whether or not to kill someone. Others will give some meaningless "rah rah" about how you can't hold ground without a 20 year old with a rifle standing there to keep it.
In response to this : first, I predict for the foreseeable future none of these fighting machines will be allowed to shoot anyone without human authorization. Requiring a human operator to directly control the machine from a safe distance away is the plan.
And second, a fleshy 20 year old is a bad way to hold ground. Robots have numerous advantages over humans. 1. Disposable. 2. Can take risks with a robot that a human wouldn't take. 3. Don't need supplies when not operating. Could deploy robots in hidden capsules located in the ground, using no fuel and minimal battery power. When something happens, months or years later, you activate the robot and guide it on it's mission. 4. A control center for an army of robots could have far more educated and experienced people manning it than the kind of people you can get to sign up for the Army and marines. Notably, you could have experienced translators, and input from high ranking officers.
Finally, robots mass produced should be cheaper than human soldiers.
Ultimately, the only thing holding this all back is technology. The KEY technology that made tele-operated robotic war-fighters impossible in the 1980s and early 1990s was that there was no way to get the kind of bandwidth needed over digital radios using un-jammable and unbreakable codes.
Notably, the communication system needed for this type of war machine is a mesh network of high bandwidth radio links (each robot would need several megabits, mostly for data from the video cameras) using electronically steered antennae to filter out jamming and allow for thousands of robots sharing the same slice of spectrum. All data would need to be communicated using a one time encryption pad.
As far as I know, the kind of radio hardware to do that was not possible before 2000, and using one time pad encryption means each bot would need to have many gigabytes of internal non-volatile storage. The tech wasn't possible in the past. It is today.
Sure, in the 1980s and 1990s there were demos of related technology, and people laughed at it and said it could never replace human beings. It can.
Note : I am in the US Army reserves as a medic.
True, but they did build Sandstorm and H1ghlander.
And, they just won the Urban Grand Challenge recently (mentioned in TFA), so it's safe to say this is up their alley.
just open it up and say "create an autonomous tank and the winner gets $14 million dollars."
Sometimes that works and sometimes it doesn't. They've had three "Challenges" now and they they still don't have a real autonomous vehicle. Just something that, on a good day, might finish a closed course.
So what they've done is actually kinda smart. They've had the Urban Challenge and identified the most promising teams, and now they're funding the first prize winner to develop a "robo-tank". Best of both worlds.
Personally, I think if they were really smart, they'd also fund Stanford and Virgina Tech, the second and third place winners. It hedges their bets and if nothing else it adds a little incentive to any future Challenges.
I think it's safe to say Carnegie Melon is focused on industry applicable results when it comes to robotics.
Maybe you should check out the NREC.
Money is the root of all evil?
The reason CMU got this funding is primarily due to the fact that we built Crusher (I'm a grad student at the Robotics Institute), for which some of this funding is directed to upgrade. Crusher is, hands-down, the biggest beast of a robot I've ever seen. It's a six wheeled, 6.5 ton, autonomous vehicle - this thing can drive up 4 foot (1.2 meter) steps, has 30 inches (76 cm) of suspension travel, and can carry 8000 lbs of payload. There isn't much that this thing can't handle.
If you have never seen Crusher in action, you've got to see it to believe it. There's a bunch of videos here: http://www.rec.ri.cmu.edu/projects/crusher/videos/index.htm.
The quote in the original post is a little misleading - I don't really think NREC is going to be working on mounting weapons on the new vehicle. Primarily they're continuing development on autonomous mobility - can it properly plan and quickly execute a good route to get from point A to point B over rough terrain. Check out the CMU press release for a little more detail on the grant.
Maybe that is what they're out to prove, but I see two major combat redundancies here, and I'm not even searching hard. And before I typed them out, I answered myself. I'll post this though because I think it will be interesting. A) If you can get a laser reading on anything with GPS, you can annihilate it via any number of GPS integrated missiles, and I'm sure the autonomous flying vehicle can do an air strike on the point too. So why not lower the lethality of the tank, and just use it more as a scout vehicle that can send valuable visual information as well as paint a target with GPS. B) You don't need much armor on the tank except to protect its engine/treads/ammunition and sensors. This thing's primary goal isn't going to be protecting lifeforms inside even though the first tank will probably be a lot like a conventional tank... For two reasons: Its easy to start with, and having a big ass tank in your lab is unfortunately worth cool points. PS: I was on the team for the first red team racing car, but all they had me do was plot some GPS points. PSS: I thought the robotic vehicle was 5-20 years in the future, not 1. LOL. PSSS: I think the ultimate combat vehicle for modern warfare that I could imagine would be a satellite up link spy tank. It could drop surveillance pods at convenient places to monitor if enemies are moving there. It would also have a few anti bomb robots it could deploy to take out things like IEDs, and to advance on the opponent where you wouldn't want to risk the whole tank. Of course, I don't think this vehicle ever should be autonomous except for uploading video and sensor information. Lets take it one step at a time, and have people safely piloting these things from a distance before jumping into the land of ED209.
God spoke to me.