Unmanned Combat Ground Vehicle

tphuang said:

funny that people would ask about dogbot, because this is happening already in large scale. So much so that they brought it to an exercise with Cambodia

Please, Log in or Register to view URLs content!

future of warfare lol. Robot dogs and drones do all the firing

and if you want to see dogbot swarm, this should be to your fancy

Please, Log in or Register to view URLs content!

Click to expand...

Current implementations in exercises are very much basic and suboptimized. I would say these are signs of initial exploratory use, but the products themselves are very much not robust nor capable for regular use and certainly aren't being used in regular TTPs.

"Happening on a large scale" is very much exaggerating current usage, more like the PLA are giving some units the okay to experiment with off the shelf products, but both the tactics and products themselves are very immature and basic. Of course, other countries pursuing this are similarly at early stages of experimentation as well, so the PLA probably aren't behind.


But in terms of the agility, stability, size, sensors, and payload of what a mature, infantry-integrated dog bot can do, current experimentation is only a barebones, small scale vision of what true capability will look like in potentially just the near future.

Blitzo said:

Current implementations in exercises are very much basic and suboptimized. I would say these are signs of initial exploratory use, but the products themselves are very much not robust nor capable for regular use and certainly aren't being used in regular TTPs.

"Happening on a large scale" is very much exaggerating current usage, more like the PLA are giving some units the okay to experiment with off the shelf products, but both the tactics and products themselves are very immature and basic. Of course, other countries pursuing this are similarly at early stages of experimentation as well, so the PLA probably aren't behind.


But in terms of the agility, stability, size, sensors, and payload of what a mature, infantry-integrated dog bot can do, current experimentation is only a barebones, small scale vision of what true capability will look like in potentially just the near future.

Click to expand...

Who are these other countries pursuing dogbot at early stages of experimentations?

The Unitree robots they are using here just came out recently and the training here are very important, since it creates data set needed to train them to be more autonomous in the future. Don't think about payload, speed and such. Think about how much more autonomous they can get

tphuang said:

Who are these other countries pursuing dogbot at early stages of experimentations?

The Unitree robots they are using here just came out recently and the training here are very important, since it creates data set needed to train them to be more autonomous in the future. Don't think about payload, speed and such. Think about how much more autonomous they can get

Click to expand...

The US, and to a lesser extent Russia, are conducting similar experiments.

Data collection is fine, but let's just not exaggerate the capabilities or the scale of their use thus far.


This just isn't really that impressive. Anyone can buy a half dozen off the shelf dog bots and jury rig a gun and a sensor onto it for exploratory experimental purposes.


However proper militarized, ruggedized, adequately armed and equipped dog bots, this is not, and certainly they are still being controlled man in the loop rather than autonomous.
So there's still a massive gap of advancements that they need to do before this can be called "large scale".


The current scale is tiny and lacking in depth, and appropriately primitive for how new this technology is.

Blitzo said:

This just isn't really that impressive. Anyone can buy a half dozen off the shelf dog bots and jury rig a gun and a sensor onto it for exploratory experimental purposes.

Click to expand...

I completely disagree with here btw. Putting guns & sensors on Unitree dogbots for experimental purposes don't mean much. The key is to actually feed it through your data models & train them to improve how autonomous they are, so you don't have to rely on controllers & things like that to maneuver around. That's something US isn't doing to the same degree.

And if they are bringing these dogs to exercises with Myanmar, then their scale may be more widespread than you think

Another UGV from Kestrel Defense.

tphuang said:

I completely disagree with here btw. Putting guns & sensors on Unitree dogbots for experimental purposes don't mean much. The key is to actually feed it through your data models & train them to improve how autonomous they are, so you don't have to rely on controllers & things like that to maneuver around. That's something US isn't doing to the same degree.

And if they are bringing these dogs to exercises with Myanmar, then their scale may be more widespread than you think

Click to expand...

I'm not sure why you would disagree.

We have no indication as to how much data they are actually collecting to train autonomy models, especially given these experiments we have seen so far seem to just involve a man in the loop controlling the dog.

Heck even if they were collecting data for future autonomy, that remains such a basic and foundational piece of research that they remain preliminary and gestational and certainly not large scale.

Taking a robot dog to a few overseas exercises for the purpose of showing off doesn't mean much -- large scale would be something like equipping every platoon in every combined arms brigade of a theater command with a robot dog at minimum.


What we are seeing right now is the equivalent of playing with remote control toys. They are only a faint vision of what the future capability will be.

Blitzo said:

However proper militarized, ruggedized, adequately armed and equipped dog bots, this is not, and certainly they are still being controlled man in the loop rather than autonomous.
So there's still a massive gap of advancements that they need to do before this can be called "large scale".

Click to expand...

Speaking of proper militarized and ruggedized. There is a video of a testing in china in which a unitree dog carries a iirc light machine gun and maintain firmly steady during a rather long burst. That's contrast to how youtube DIY gun dog which would flinch and need to readjust every time it fires. That definitely looks like factory level support for me.

PS: can't find the original video mentioned above but after checking newsreel of recent Cambodian exercise it seems to be two different model. There is off the shelf unitree dog weighs 15kg and only do recon mission. Then there is this "B1 quadruped robot dog" weighs more than 50kg carrying a refile with buffer mechanism and this thing wouldn't even flinch when firing.

kriss said:

Speaking of proper militarized and ruggedized. There is a video of a testing in china in which a unitree dog carries a iirc light machine gun and maintain firmly steady during a rather long burst. That's contrast to how youtube DIY gun dog which would flinch and need to readjust every time it fires. That definitely looks like factory level support for me.

PS: can't find the original video mentioned above but after checking newsreel of recent Cambodian exercise it seems to be two different model. There is off the shelf unitree dog weighs 15kg and only do recon mission. Then there is this "B1 quadruped robot dog" weighs more than 50kg carrying a refile with buffer mechanism and this thing wouldn't even flinch when firing.

Click to expand...

The ability to mount a proper weapon (like an assault rifle or a LMG) as well as an integrated sensor system, all in a relatively accurate mount that is stabilized, will require a fairly large robot dog.

Many of the experimental uses of robot dogs around the world have involved mostly knee height robots at normal operating height, which is fine for tight spaces and acting as a forward node/recce in urban environments I'm sure.






.... however a proper, ruggedized full scale squadmate robot dog that is armed and equipped with sensors with good endurance, may be something that is larger that is shoulder height at regular operations.
Something like the original visions for the Bigdog and so on, but with the benefits of modern robotics and motors -- think the likes of Unitree B1 or Boston dynamics spot but two or three times enlarged in all dimensions appropriately. Something that size equipped with say, an intermediate cartridge gun (or alternative weapons that accommodate a similar weight and profile), and a TV+FLIR sensor, both fully stabilized, is something I could see a robot dog "squadmate" as being in the future.

Blitzo said:

I'm not sure why you would disagree.

We have no indication as to how much data they are actually collecting to train autonomy models, especially given these experiments we have seen so far seem to just involve a man in the loop controlling the dog.

Heck even if they were collecting data for future autonomy, that remains such a basic and foundational piece of research that they remain preliminary and gestational and certainly not large scale.

Taking a robot dog to a few overseas exercises for the purpose of showing off doesn't mean much -- large scale would be something like equipping every platoon in every combined arms brigade of a theater command with a robot dog at minimum.


What we are seeing right now is the equivalent of playing with remote control toys. They are only a faint vision of what the future capability will be.

Click to expand...

This is the beginning phase of what I see have already happened in ADAS industry.

Have you taken a look at Unitree Go2? if you haven't, then you should
It uses advanced 4D Lidar tech that significantly improves IDing of objects

This is pretty far ahead of the sensors that Boston Dynamics Spot uses (which I assume is the best that US army could work with)
which is basically IMU + vision
That's imo just pure garbage
It also got in with Hydraulics drive robot at that early stage and have now realized that doesn't work and is finally moving to electric drive
This is all emblematic at a subpar system developed at great cost at a time when autonomous tech simply was not ready for market yet.
And that's what US army is stuck with.

Go 2 only came on unveiled in July & PLA already has it in oversea exercises. We've also seen them train Go 2 with drones carrying them and transporting them to destination.
That's pretty fast progress for a few month

Unitree itself talks about AI & Robot GPT used to make decisions, which over time through field deployment is how you collect data to continue to refine the model

So given what I've seen just in a few months, I can extrapolate this over a few years and see them mass produce a bunch of dogs that operate autonomously.

That's just not possible with US military & Boston Dynamics. It's like watching Mobileye in action in ADAS. An earlier player stuck with archaic technology & enable to move forward. Also built a system that is too costly to see mass adoption in civilian field, which means they won't be able to scale up production.

Whereas for Unitree, when you price something like $1600, you know it's going to see a whole bunch of demand around the world and see huge vertical scaling, enabling robot dog army in the future.

When I posted these clips in response to robot dog army, it wasn't about what they have now, but what will they have. The key here is that they have the domestic firms here with commercializable product that is clearly moving very fast. I could be jumping the gun here in thinking they will get to a certain point in say 5 years, but what I've seen in EV and drones industry tells me that I'm not.

I want to point out here that the significance of employing Unitree Robot Dog here isn't it's current payload or being able to carry a gun, but the commercial viability of the platform, the massive amount of data that it will gather which points to mass scaling and becoming mostly autonomous.

In my mind, having a mass army of robot dogs that can operate autonomously and be air dropped behind enemy lines with drones & operate in sync with drones after that is more important than having a large robot dog that can carry more stuff. You can replace latter with tracked UGVs we've already seen. Cool part of robot dogs + drones is that they would be hard for radar system to detect just due to how low and slow they can fly & being much smaller than helicopters