Home » News » Army » DARPA shows off Big Dog ground drone to USMC Commandant

DARPA shows off Big Dog ground drone to USMC Commandant

by Mike Hoffman on September 10, 2012

The Marine Corps Commandant and DARPA’s director got a first hand look Monday at what DARPA used to call the BigDog ground drone and is now calling the Legged Squad Support System (LS3).

Boston Dynamics has built the ground drone. DARPA officails developed the LS3 to give soldiers’ and Marines’ backs a rest. The legged drone is designed to carry 400 pounds and follow a squad leader over rocky terrain, up hills and through brush.

LS3 had its first outdoor test in February when DARPA released this video of its performance.

Since February, Boston Dynamics engineers have built LS3 to run quieter and transition between it’s trot, jog and run modes.

“LS3 is now roughly 10 times quieter than when the platform first came online, so squad members can carry on a conversation right next to it, which was difficult before,”Army Lt. Col. Joe Hitt, DARPA program manager said in a statement.

The ground drone can trot through rocky terrain at 1–3 miles per hour. It can jog at 5 miles per hour and then run at 7 miles per hour over flat surfaces.

If the BigDog…err…LS3 is knocked over, the drone is programmed to stand back up. It also doesn’t require a soldier or Marine to drive it. It will naturally follow a leader.

“The vision for LS3 is to combine the capabilities of a pack mule with the intelligence of a trained animal,” Hitt said in a statement.

The Marine Corps started a 2-year program to develop the LS3 in July. The first test hosted by DARPA and the Marine Corps Warfighting Laboratory is scheduled for December.

Last point: Why change the name from BigDog to LS3? Is there a Pentagon acquisition rule that states you must have at least a four-word acronym that includes the word system if a service wants to buy a weapon or vehicle? I don’t get it.

Share |

{ 75 comments… read them below or add one }

blight_ September 10, 2012 at 4:34 pm

Need one that can tow a mortar. Not optimistic about firing one from big dog…


AnonDog September 13, 2012 at 9:03 pm

Look up Alpha Dog.


Taylor September 10, 2012 at 4:40 pm

Seems like a mule would be cheaper.


Anon September 10, 2012 at 5:17 pm

I had the chance to run that by some folks on the DSB a couple times. I pointed out that a mule with blinders, earplugs, and a little operant conditioning could do the same thing, except it would be mostly self-guiding and could "autonomously refuel with field-sourced resources". The reply was… less than favorable.


Dan Gao September 11, 2012 at 12:00 pm

That's like saying a carrier pidgeon is "cheaper than UAVs". It's a knee-jerk and rather ignorant comparison. Can a mule take a burst of 7.62mm rounds and keep moving? Which do you think gets tired faster or can carry more? Does a robot get scared by loud noises or explosions? How often do you have to feed an animal vs. refueling a machine? You have to look at the bigger picture, and also the longer run. Whatever limitations this might have now, think of in five, ten, or fifteen years. Maybe it will be able to run 30 miles per hour over mountain terrain, drop kicking bad guys in the nuts before they even know what's going on.


Hubble September 11, 2012 at 4:26 pm

How about putting a bullet resistant vest on a blind / deaf mule. But yeah, I get your point. These things are in the development stage and aren't costing billions to develop…yet. So far the cost is in the millions. Keeping 50,000 bulletproof vest wearing mules ready for battle duty costs more than that.


elmondohummus September 14, 2012 at 2:58 pm

You'd have to also take away it's voicebox so it couldn't bray and give away its team's position.

On top of that, what's the solution for dealing with its waste dumps? A human can "hold" theirs, so to speak, and deposit it where its difficult to be found. Can mules be – for lack of a better term – "housebroken"? Or would the solution be to designate some poor soldier to "unit pack animal policing duty" with a broom and wheeled trashcan?


Before anyone takes this as condemnation of Anon's post, don't read too much into that. I too have wondered why the Army and Marines don't resort to horses – not mules, not in my musings – for certain applications. I figured they'd be excellent in some of the terrain in Afghanistan; not really mountainous terrain of course – a horse can't climb a sheer wall – but restricted areas where vehicles have trouble going and helicopters have problems accessing. Seems to me that horses and pack mules would be a perfect solution to increase unit endurance… (cont'd…)


elmondohummus September 14, 2012 at 2:58 pm

… cont'd:

That said, no one who's read the history of the military should think that simply going back to those would be a panacea. Forage is being only given lip service in this scenario; "autonomous refueling" could only happen when 1. It's something that the horse *would* eat (although thankfully, I don't think they're super picky), and 2. When sufficient quantities exist to feed that horse. I'd imagine it'd be hard for a horse to forage in, say, the rocky terrain areas of Afghanistan. Remember: Caring for horses was a HUGE problem for the WWI army, and they would've known about letting those guys eat the grass wherever they were at. It's obviously not as simple as letting those things eat the local plant life; if it were, that's what they would've done back then. (Cont'd…)


elmondohummus September 14, 2012 at 2:58 pm

… cont'd:

On top of that, pack animals are not low-maintenance systems. They're simply different types of care. A mule can follow, but it can also have a mood to not want to work at a given time. Ditto a horse, and they both can be spooked, too. At least a vehicle can be turned off and left that way when quiet is needed; how do you shut a panicked mule or horse up short of shooting it? Furthermore, while both vehicles and pack animals can operate "injured/damaged", a vehicle can be repaired. Good luck "fixing" a mule.

It's not like pack animals should be automatically dismissed as a low-tech idea. I do agree with anon there: Sometimes a bureaucracy will reject things out-of-hand without due consideration. On the other hand, mules were evolved out of the military for some reason. Those reasons can't be ignored.

Curt September 11, 2012 at 8:17 am

It seems that way, but think about it.
1. A mule has to eat, even when not working. So although it can graze, you need land for them to graze on, vets to check on them, trainers and handlers to work with them, breeding stock, etc. The robot just sits quietly on a shelf until needed and when you're done with it, you can recycle it.
2. A mule needs to rest, a robot just needs more fuel.
3. A mule would need to be trained to enter a truck, helo, be parachuted, fast roped, etc. (OK, good luck on all those). A robot can potentially be slung under a helo, stacked on the back of an APC, and HALO or fastrope with the best of them.
4. Can a robot charge the squads electronics? Maybe if they use a methane fuelcell but collection is a problem. Just plug in to the robot.
5. Mules need a separate and bulking logistics train (hay and oats) in areas with limited forage (like say Iraq or AFPAK).

So while it seems a mule is cheaper, in reality, it isn't even close.


shawn1999 September 11, 2012 at 9:38 am

Don't forget- when not in use, the Robot gets turned off and costs only storage. A Mule still has to be fed, walked, cleaned up after, etc.


Riceball September 11, 2012 at 11:05 am

But one benefit of a flesh and blood mule is that in a pinch you can eat it; can't eat a robodog. Granted it's not very likely that we'd ever be in a situation so dire that we'd have resort to eating pack animals but it can happeen. Of course, having a mule just so can eat it in a pinch is hardly a reason for not going with a robodog, if we thought like that we wouldn't have gotten rid of all our pack animals in favor of trucks.


Mastro September 11, 2012 at 4:23 pm

It helps that you can let a mule die (not enough food, let it die, too loud? let it die, tired, etc)

Leaving behind a robot for the Chinese to reverse engineer might not be what we want to do. Common sense tells you that this would happen in the first year of use (its not a plane- any malfunctions- we have to leave it) – so basically we are spending million to do initial engineering on the next Chinese /Russian battle dog.


Curt September 12, 2012 at 10:27 am

But a robomule doesn't try to kick or bite you either so you don't want to eat them!


elmondohummus September 14, 2012 at 3:11 pm

Well, you'd just have to build the robodog out of the right stuff to eat one…




Thomas L. Nielsen September 12, 2012 at 3:14 am

"A mule would need to be trained to enter a truck, helo, be parachuted, fast roped"….

I just had this image go through my head of a mule screaming "GERONIMO" as it gallops* off the rear ramp of a C-130….

Regards $ all,

Thomas L. Nielsen

*) Do mules gallop? I have no idea.


elmondohummus September 14, 2012 at 3:00 pm

Aw, hell, you beat me to it. By 3 days. :(

Guess I should read the entire comments thread before posting next time.


Johnny Ranger September 10, 2012 at 4:43 pm

That is the coolest thing ever.


Andy September 10, 2012 at 4:44 pm

Need to have a Machine gun and Camera mounting in front and have the DOG taking a LEAD….


Noha307 September 10, 2012 at 4:52 pm

Commandant: How is the F-35B program coming along?
Adjutant: Well, sir it…uhh…um… Hey, look, a robot dog!
Commandant: Oooo, cool! What were we talking about again?

At least it no longer sounds like (and is as loud as) a leaf blower.

(In all honesty though, these things are pretty cool.)


jrrr September 10, 2012 at 5:26 pm

1) Why does it pull its feet up so quickly? It looks like it's constantly trying to unstick its feet from tangles. Perhaps this motion makes it harder to tangle in the first place?

2) Not sure how to word this – why does it move its feet so frequently? It appears programmed to maintain a constant "stride rate" regardless of ground speed. This is particularly apparent at around 1:17, when one machine is stationary but stomping, and the other is moving very slowly but stomping rather quickly.

On both points, it seems like the natural stride of an animal would also be more efficient for a machine.


Stormcharger September 11, 2012 at 8:39 am

It has to do with how it balances. As soon as one foot comes off the ground, it begins to fall over, keeping the feet moving quickly allows it to stay upright. If the stride rate changed, some other means of affecting balance would need to be used since it doesn't have arms and would only serve to add more weight to the vehicle. By maintaining the same stride rate, the only variable to how fast it moves is then length of stride.


jrrr September 11, 2012 at 9:29 am

That's a good point – a big difference between bigdog and other ambulatory robots (or animals, for that matter) is that it doesn't seem to have any movable joints other than its legs, giving it fewer controls over the balance equation.

Perhaps a future version could incorporate a few joints in the body for this purpose.. or perhaps they've researched this and determined that it's more efficient to stomp around like a hyperactive child =]


Raraavis September 11, 2012 at 10:51 am

When I saw the added pieces which I guess is it's rollover cage. I was hoping those were articulated joints. We are going to have to add many more joints to these things if we ever want to increase their agility. At this point this is just a car on legs.


Stormcharger September 11, 2012 at 11:49 am

It also shows just how complex the structure of humans and animals really are. The daunting technical hurdle is the programming needed just to maintain balance and move around on legs. If random chance and primordial ooze can do it, surely a couple guys at MIT should be able to figure it out… Just saying.


Raraavis September 12, 2012 at 6:12 pm

are we giving them the same time frame the primordial ooze got?

Roy Smith September 10, 2012 at 5:36 pm

They're still trying to do something about the fart noises it makes while moving.


Glen September 10, 2012 at 6:16 pm

What happens when…
- fuel tank / carried ammo catches fire and it keeps "following the leader"?
- someone shoots a hydralic line / joint / fan / etc?

Seems a track vehicle would be better suited…


jamFRIDGE September 10, 2012 at 6:48 pm

I imagine they’ll put some sort of protective armor, be it angled or Just hanging kevlar vests. Just my thoughts


shawn1999 September 11, 2012 at 9:48 am

Shoot one or two knee joints. It'll fall over easily since it has to keep stepping to maintain its balance. Then the soldiers will have ~400 pounds of gear to unload, during which time they are sitting ducks. And, if they trained with it, they won't be used to carrying their own gear, so those who are left will tire easily and can be engaged at a later time after they have fatigued.


Kevin September 11, 2012 at 10:44 am

If the enemy is close enough(or a good enough shot) to take out "one or two knee joints" then you've got bigger problems.


Raraavis September 11, 2012 at 10:59 am

I agree if you can get a good enough shot to hit the mules knee than a better use of that opportunity might be a head shot on the squad leader.

The third generation of these dogs are going to be the soldiers not their mules. Then it is going to be a bad day to be a human combatant or a dissident or a protester or someone with a opinion that disagrees with the government.


Hikerguy September 14, 2012 at 12:06 pm

In that case these will be the "Cylons" of the future.

snork September 11, 2012 at 12:09 pm

And what about those wheels on supply tucks??? What if the enemy sneaks up and slashes the tires!!

Armchair tacticians…


Riceball September 10, 2012 at 7:05 pm

It is still just an early prototype and clearly a work in progress. I'm betting that the final thing, if they decide to actually take the project to production, won't look much like this prototype does and will probably have some light armor on it.


tmb2 September 10, 2012 at 8:04 pm

DARPA is like the comedy club where Saturday Night Live superstars got their start. Don't look at any DARPA project with full-rate production in mind.


Curt September 12, 2012 at 10:36 am

This is for infantry in difficult terrain where you can't get a vehicle but it is still passible by troops. If you can get a tracked or wheeled vehicle there, it doesn't make much sense to send a walking robot.

As for following the leader, it probably will have a variety of modes. Just off the top of my head:
Robot, follow! Follows the person with the controller or who gave the command
Robot, sit! Stay in place waiting for the next command
Robot, sleep! Goes into storage mode
Robot, charge! To draw fire or create a distraction
Robot, go to Joe! to deliver equipment and/or recover injured personnel

So if it catches fire, send it away. You could even send it toward the enemy! As opposed to a mule that will be a hazard to anyone around it.


orly? September 10, 2012 at 7:03 pm

How many versions of an unmanned ground supply platform do we need?


Matt September 10, 2012 at 7:15 pm

Kinda loud, no? Would be OK I guess for long insertions or patrols but eventually you have to turn the damn thing off to hear!

Would be great to have it follow you to a staging area, shut it down, then go on the mission… then whistle twice and powers up and appears out of nowhere with water bottles, ammo, medical, etc. It comes to you like ol' Paint used to do in the westerns.

Ah…. now I see it.


Ron September 10, 2012 at 8:01 pm

What happens when it see's a cat?


Russell Romick September 10, 2012 at 8:34 pm

Haha I was thinking the same thing, or a car or a dog running past its field of view, does it try to follow? I would think they will eventually have some command conditions like follow only the guy with a chemlight on his helmet or something similar. I understand the need for research and prototypes to advance technology but really, there are humvees, firescouts, quads, deuce and a half's, motorcycles, and yes mules. Spending money for something you don't need nor will you ever is so wasteful. Let the Japanes continue designing their robots. Buy a bunch and have them as "porters" they are nearly as silent as a human(electric) and can go in almost the same terrrain. I understand the convenience and redundancy of a 4 legged version but the cost will spiral out of control like every other defense project. Everyone think oh yeah the LCS was outrageously expensive but we have learned our lesson and so has the contractor……nope. Every project in the past 20 years has spiraled out of control to the point of full reviews being done and some construction temporarily halted and charges discussed. And you know what we do every time? We always ALWAYS ALWAYS pay, and pay through the nose. The contractors are always laughing at us as dumb slow-witted taxpayers. We have such horendous fiscal mismanagement it is stupid. I first noticed it at the time of the tanker scandal but it was before and since, and until we wake up and say okay you say 1 billion? 1 billion it is….no cost plus, no restructuring. That is it. Oh now you say 1.2billion? Okay you, you, you go to jail and now we own 50% of your company. Fraud is fraud. CAn you tell I am fed up with this crap?!?


Rob September 10, 2012 at 9:25 pm

The drone is amazing achievement. Soldiers aren't as beefy as they use to be, having something help them carry extra equipment would be useful.

However, wonder how they'll sort out endurance. MAINTENANCE in the Field.. More High tech the plumming, easier it is for drain to get clogged up.

Where heck they going use this thing? I hope the soldiers in the field will be able make sense in putting these expensive machines before production is approved.

As side note, i don't like fact their not calling them Big Dogs anymore.


Matt Holzmann September 11, 2012 at 4:53 pm

soldiers are carrying 30-40 lbs more gear than they ever have in the past. They are breaking down humping all of that gear up and down mountains at altitude. Their conditioning is better than ever, but the grind is the grind.

Better to import some mules and muleteers like we used in Italy in WW II.


MilGuy September 17, 2012 at 10:00 pm

The idea that today's military is somehow less physically capable than any time in history is pretty laughable.


dubweiser101 September 10, 2012 at 10:44 pm

Will it play fetch and bring me my slippers and paper in the morning?


AAK September 11, 2012 at 1:53 am

I like these creepy things and in 10 years I’d imagine it;ll be a whole lot slicker. But for virtually any terrain this is supposed to operate on I think a flexible-tracked vehicle with a self righting mechanism would be just as effective and a whole lot simpler.


jghj September 11, 2012 at 9:47 am

just get a fucking horse


Dan Gao September 11, 2012 at 12:04 pm

Yeah! Who needs radios? Just get a fucking smok signal


Kevin September 13, 2012 at 5:28 pm



Arthur Russell September 11, 2012 at 10:19 am


How much weaponry could you load on it? Add in remote control or ai targetting, You have a “mini-tank” if armored. Then add in some “active defense”. How long before you no longer really need a guy to “lead” the thing?

400 lbs? 100 lbs armor-300 lbs-remote controlled minigun? Now upsize it. Remember, this is a prototype.


PolicyWonk September 11, 2012 at 11:23 am

While impressive – this thing is awfully loud (granted, It is a prototype).

What concerns me, is that even though it'll carry 400 lbs now, it somehow seems unlikely the load marines are physically carrying will get any lighter.


IknowIT September 11, 2012 at 1:26 pm

What is the benefit of this, in total, over a real mule?


Raraavis September 11, 2012 at 2:48 pm

In 20 years a mule will still be a mule, while in 20 years the BigDog will be a ultra-fast ultra-powerful semi-autonomous killing machine that can be used to fight wars abroad and suppress any descent at home.


Ray September 13, 2012 at 9:44 pm

In my mind, the main benefit is in peacetime. You can load thousands of these in crates and store them for years with little or no maintenance until they are needed.

To have thousands of mules available, you need to keep a standing army of mules. You also have to feed the mules when not using them in war and employ people to care for and breed the mules.


Tribulationtime September 11, 2012 at 2:40 pm

"IT" would grew in something usefull. Right Now I gonna keep one eye on my pet f***ing around my mower


JohnB September 11, 2012 at 10:42 pm

Super rich men 's war toys.


anon September 12, 2012 at 3:56 am

All in all, 's a Good Idea, though the road to hell is paved with the best of intentions. Definitely not feasible in the short-term future. *shrug*

China's response: smaller versions that carry less, are cheaper, and by the billions. And they explode (command-det or accidental)


Rinkusu September 12, 2012 at 5:14 am

400 pounds, hmm… So-o-o if we mount soldier on top of it, give him mg or at weapon we would have mech cavalry unit.
Not to mention mech cowboy potential in Hollywood!


blight_ September 12, 2012 at 11:28 am

Robot horse cavalry, for the day we need to ride down those foot-mobile Taliban.


Joe America September 12, 2012 at 2:35 pm

It follows the leader….
So what happens if the leader ends being a bad guy?


DanS September 12, 2012 at 4:08 pm

Man is that thing loud.


traindodger September 12, 2012 at 4:09 pm

I've said it before and I'll say it again: the psychological warfare potential of a platform like this is limitless. The creep-out factor alone is overwhelming. Plus, this is just the first generation of the technology. Maybe ten or twenty years from now, these things will be faster than the fastest Olympic sprinter, agile, armed to the teeth and practically silent. This is just the beginning.


jeff September 13, 2012 at 4:54 pm

remember we had mounted cav until we had jeeps and we did not loose the last mule until after Korea. not to mention in Vietnam the used bicycles to all huge amount of cargo


Hefe September 15, 2012 at 2:43 pm

I am a bit frustrated that darpa gets millions of dollars to build a gigantic robot, but the military still hasn't built or bought a more dependable firearm for our grunts in the field. We know our soilders fire guns that's fact, the robot is theoretical. Seems the military needs some better prioritites.


Timer4ever September 17, 2012 at 8:47 am

Things are getting really spooky out there. We love to bring the kitchen sink with us in combat. Just remembering Charlie and his baggie of rice.


Dean September 18, 2012 at 9:40 am

So many people are sceptical about this project , im surprsed the potential applicatioins for this and knowledge gained in the development are priceless this could really help to win battles and save lives , well done Boston Dynamics


Guest September 21, 2012 at 4:41 pm

Ever since the failure of programs like FCS, trashing whatever high tech programs the military is working on has been "in vogue" among the armchair warriors.


paul September 22, 2012 at 11:00 am

looks like a total waste of time and money in its current form to me!
yeah , toys are neat but I can,t see this as anything but a money pit.
it looks far too noisy, slow and vulnerable to rifle fire , and I can,t see it as an asset


Make Money Online December 12, 2012 at 2:19 am
dishwasher parts kitchenaid April 25, 2013 at 8:42 pm
Anon September 10, 2012 at 11:05 pm

You're right about the upper limit issue. Plus the BigDog is foundational gait and balance research with plenty more applications. When it comes to short term solutions though, and speaking in general, I've been frustrated with the fact that the defense research system is set up to give money only to new, flashy solutions. It's a pattern I keep seeing crop up: low tech gets rejected, even when it's clearly right for the job and cheaper than anything else will be for at least a decade, because nobody wants to fund it. It's natural selection, or rather, bureaucratic selection. The emergent behavior of the funding system is a selection pressure favoring complicated, hard solutions, and killing off the easy ones, so everything gets expensive and bloated. DARPA is particularly bad at this. With "Advanced" built so deeply into their mandate, anyone who finds an easy, low-tech solution to a DARPA problem loses their contract because it no longer fits DARPA's mission. Maybe we could benefit from an analogous group dedicated to finding simpler, cheaper ways to achieve the mission. It'd be tough to keep it going in this rooted culture of rice-bowl-filling, though.


Thomas L. Nielsen September 12, 2012 at 3:19 am

A genetically engineered, cyborg mule? Why do I get this image in my head of a mad scientist type in a crypt somewhere going "They said I was mad! MAD! This will show them. MUHAHAHAHAHAHAHAHA……"?

Regards & all,

Thomas L. Nielsen


Curt September 12, 2012 at 10:20 am

You could even build in a small explosive charge to destroy the computer and critical sensors. But even then, it is more important to protect the production knowledge as opposed to the actual pieces themselves. The Iranians have had an F-14 for years, but they can't build them because they don't have the production knowledge.


blight_ September 13, 2012 at 11:06 am

Unlikely to have a self-destruct in it.

More likely to have a GPS coord, and if we can spare aircraft or a UAV, a Hellfire will come down in a few hours.


Aya November 18, 2012 at 10:37 am

One important thing tguhoh is that the developers have to realize what are they testing. The real-transport test cases are not a substitute for integration tests against a real application instance (unless the application on the other side is so simple that the mock can replicate the full functionality in the test case). In the described case, the incremental coverage over the VM transport tests is that we are testing that the Mule transports are not introducing unexpected side-effects.What we are NOT testing is that the ESB application would work against the real points of integration. I.e. the JMS communication might be fine, but the actual data might be wrong.


Leave a Comment

Previous post:

Next post: