DARPA code teams compete on same K9 robot

News 42 Report

""walking becomes more like controlled falling"..... into Quantum Leaping. AI Mentored Space.

I Kid U Not.

Nice approach

Just as long as they realise that we do fall over occasionally.

Might be a dumb question...

I remember dogs, I remember watching them move, I also vaguely remember reading something in your article about the software being a bag of shite.

One question. Why does only one leg move at a time?

From my recollection of dogs, they move two opposing legs at the same time, and get past the balance issue by allowing each leg to move at speed enough so the falling of the body in the direction of the current position of the centre of gravity is propped up by the speed at which it moves, while the centre of gravity shifts toward the end of the movement.

Call me silly, but surely if you want to mimic something, then actually think about why what you're mimicking moves in the way it does. If they up the speed of the legs, and have the async switcheroo that a dog does, then they'd get some speed out of it, up until the point where a bounding stride is required (both legs of front/back move synchronously in opposing directions, unlike a horse running). Terrain issues seem only limited by the 'dogs' reliance on ensuring each individual movement is completed before the next can begin, it seems that the dog needs too much thinking time for doing something which should be controlled quite easily.

The next question is... Why are roboticists so crap? It seems that the design decisions went something like this...

Engineer 1: Why don't we make it move two legs at once, y'know like a real dog?

Engineer 2 (senior position): Nope, we tried that back in '73 with a larger [heavier] model, the CoG kept shifting too far and made the machine unsteady, so we [still? with todays gyroscopes?] need three points of contact to make an effective 4 legged walker.

I remember one of the asimo developers saying something like, the trick with asimo that took us 5 years to learn, is that human beings are essentially propping themselves up while falling with each stride. Thats why asimo is so realistic.

Uh...

"NewScientist reports that most of the groups have now moved away from stable walking, in which the LittleDog keeps three feet on the ground, to "dynamic" motion more like that which animals and people use."

ROTM avoided?

Forget armed milbots, embrace the robohound humping your leg :D

WHY?

How much money? .. call me picky but why not use real dogs? they can already do all these things, cost bugger all to make and don't need petrol. With the added bonus for the future squaddie that, when he arrives at the location, he has a fresh meat for the pot! kinda like killing 2 dogs with one stone I would say!

Horse?

If you want somthing to move like a horse, and think like a horse.... Even smell like a horse... Why not just use a horse.

Whats the point!

@Why

Really now, use some thought.

Why not use real dogs? Lets see... Real dogs require training, the results of which are highly non-standard, based upon the trainer's skill, the dog’s aptitude, and a myriad of other factors. Then, there's the fact that dogs are thinking animals, and, even when well-trained, will not always do what you want, instead choosing to do what *they* want. Then you have simple failures to communicate - happens all the time, even when you have two humans talking face-to-face; anything that removes the ‘interpretation of directions’ issue from the question is a big 'plus' to reliability. Dogs require dedicated handlers, people whom might otherwise be pointing a rifle. The droids can be operated by anyone whom has been through a fairly straightforward orientation, meaning that *every* rifle-pointer is also a potential droid handler. Then there's the fact that when you're not using them, dogs still need to be fed and cared for, whilst the droid can be put in a crate and forgotten until next time. Dog grow old fairly fast, and need constant replacement, whilst the shelf life on the droids is presumably quite lengthy. An injured dog is out of action, but the droids can be repaired fairly close to the front line, and possibly even on the line itself, give some basic knowledge and the right spare parts - spares which might well be cannibalized off another droid. Can't swap limbs on wounded dogs... not in the field, anyway!

The lists of reasons go on and on. It requires almost zero critical thought to come up with valid reasons why this may be a practical idea; the list above came off the top of my head as fast as I could type. I'm quite certain I haven't hit all the advantages yet, either.

The point is to get the utility of a pack dog, with none of the draws-back of live animals.

I prefer the Dr Who version of K9

Now that "dog" was cool, and pretty effective against all sorts of nasties.

Title

That walks like a chameleon not a dog.

Can i get one to climb up and clean the leaves out of my gutter?

Not even near to K9

where's the flamethrower for a nose????

Wrong

A dog of that size would jump from flat spot to flat spot not try to walk over it thats utterly undog like. As far as I know dogs suck at climbing anyway try the goat for a model.

Horses.

".......Why not just use a horse."

'Cos courtesy of Operation Barbarossa we know that you end up in a huge logistical pickle where eight out of every ten horse-drawn transports carry horse fodder. This means that you can't get sufficient supplies to the front and you fail to capture Moscow before winter sets in.

TeeCee

Why you don't want real animals on a battlefield

Especially not dogs - they can bark at the most inappropriate times (you won't sneak anywhere, or remain concealed, with a barking dog at your side). Aside from that small inconvenience, it probably takes a lot of training to not have an animal run in terror when the shelling starts.

And finally, soldiers will have a hard enough time coping with the deaths of their fellows. Throwing a live animal in that kind of hell will probably add to the psychological misery and unfairness of the battlefield. I mean, soldier or not, people have a tendency of getting attached to animals, all the more ones that help them.

Seeing the pack dog get blown away will not help morale.

Seeing the pack dog get blown away will not help morale.

Wimps. Polar explorers have never had a problem with it. (I live in Norway)

And dogs are easily debarked.

>whilst the shelf life on the droids is presumably quite >lengthy.

You surely jest. If it's more than a year old it will be incompatible with the latest patches, the spare parts will be unavailable. After all these things are not likely to be mass market devices, you won't be able to nip into PC World for a spare nose or left hind leg (and they won't service it if you aren't running Windows anyway!).

>but the droids can be repaired fairly close to the front line,

>and possibly even on the line itself, give some basic

>knowledge and the right spare parts

Can a soldier repair his tactical radio? Can he even fix his rifle?

I suspect that the enthusiasts for the 'high' tech approach are either very young or have been hidden away in academia their whole lives and have never had to deal with real life computer technology.

@kevin

Wimpiness has nothing to do with it, and your misinterpretation as being an issue of manliness is amusing. Dogs do indeed work better in extreme cold, snow, and ice, but how much fighting do you think is actually going to be happening in the arctic, eh? Since the Winter War(s), there's been damned little of it, really. These robots aren't going to show up on Robert Sorlie's or Jeff King's teams anytime soon, but then they were never *intended* to, either.

Considering that military hardware often sits on the shelf for *decades,* I'll have to dismiss your 'obsolescence' suggestion out of hand. You clearly don't understand military procurement cycles, nor do you understand logistics purchasing. The military fields almost *nothing* that is truly cutting edge - almost everything they field is behind the curve, but it *is* tested, solid, reliable (for the most part). The M1 tank design comes from the 70s, the F-35 design started more than 15 years ago, the radios that the soldiers carry are decades in the design, and so on. Once the bugs are worked out, there'll be very little 'updating' going on, and any updates that do happen will be written by poor bastiches whom have to labor under a specification that requires compatability with existing systems.

As for repair in the field, what do you think soldiers do right now with their mechanical systems? Eh? fix, patch, modify. What do you think is going to be the main points of failure and vulnerability on the dog-bots, anyway? the elctronics? Not likely. It'll be the mechanical systems, especially the leg servos. Any moderately experinced tinkerer can fix those.

I'm hardly young, and I spent a career in the belly of the miltary beast, and I'm still well-connected to said beast by friends and relatives whom serve in multiple areas and forces. These dog-bots are *far* from ready for prime-time, but they have real potential, if they can be made to work at all.

@laird cummings

The wimp comment was directed at the morale issue not at whether or not they would be well suited to arctic conditions. Nor did it have anything to do with the location of any future war (although the Russians do keep irritating Norway with exercises and flybys it doesn't mean I believe they will be marching over the border).

>The military fields almost *nothing* that is truly cutting edge - almost everything they field is behind the curve,

Surely that rather supports my point. In order to do what is claimed for them such devices must be cutting edge, at least as seen from now. So either they will be useful and vulnerable and deployed soon or robust and not very clever and deployed later.

Look at what happens with high tech equipment like modern helicopters.

I'm fully aware that the modern military tends to field only equipment that has been tested over long periods, etc. Which is one reason why I am extremely sceptical of the enthusiasm for this kind of automation.

Also, it is another technological, tactical, solution to what is in fact a strategic social problem.

Not to mention...

You can dump a set of GPS coordinates into a packbot, and it'll walk to it. I don't know of any dog or horse that can be told to go to a spot it has never been to before.

@Kevin, again

Actually, your point isn't supported at all - the enthusiasm is for a technology that, when fully developed, will be of very high utility. Simply because it's not ready to go *now,* nor will be anytime soon, is no reason to not be excited about it. Enthusiasm now carries the development forward. In five, or ten, years these will be standard devices, and well worth the time and effort.

OK, it'd be really damn nice to have them *now,* but I understand these development cycles very well, and look forward to the end result - Miracles are not looked for, nor anticipated. I anticipate results that will stand up to war, and that takes time, but you must start *somewhere,* and this IS that somewhere. Of course I'm excited! This will be a major advance in warfighting that address the soldiers' basic needs in a realistic fashion. Or it will be, once it matures.