dan/california

155, GMLRS, ATACMS. Bradleys. All of these have been proven to work, they just need a ton more of them. NASSAMS and missiles for it. Then put real money and technical effort behind Ukraines drone programs. Last but most certainly not least  would be more training, preferably measured in months, not weeks.
 

155, GMLRS, ATACMS. Bradleys. All of these have been proven to work, they just need a ton more of them. NASSAMS and missiles for it. Then put real money and technical effort behind Ukraines drone programs. Last but most certainly not least  would be more training, preferably measured in months, not weeks.
 

As @chrisl noted, the terms autonomous and automated are getting increasingly murky these days and pretty much depend on context.
Your definition of autonomous, "no communication with a human" would also fit a system where e.g. a room has lots of cameras hanging from walls and ceiling, connected to a computer. The computer controls a bunch of miniature or real cars which themselves are totally stupid. Both are existing scenarios, so the definition is not wrong. But probably not what we would mean or want in the context of drone warfare because you still have the line of communication with the controller who just happens to not be a human.
Then there is edge computing where you have sensors in the vehicle and do some lightweight (pre-) computing there but rely on a central server for everything else. Probably also not what we want.
In addition you can have fully independent systems that gain additional capabilities by communicating with each other or (centralized) infrastructure. Think autonomous vehicles that can do cooperative maneuvers that would otherwise require LOS, for instance.
In autonomous driving we would usually want fully independent systems but that is more for safety considerations.
A real drone swarm is a totally different concept, I think. Of course it could fit the "classic" definition "independent but with (sort range) communication among all the swarm members". But one could also imagine a single system, as you suggest, that consists of many parts and distributes computing etc. between the different members. That would have really interesting features as it could become more "intelligent" the more members (and thus computing power) it has and so dynamically do more or less complicated stuff.

For those who remember the British 'Dragonfire' anti-drone laser test from January, Grant Shapps (UK defence secretary) is now talking about possibly delivering it to Ukraine relatively soon.
It's timeline was originally aiming to be in service 2032 (assuming it can be made to work adequately). The time line was accelerated to 2027, because I'm sure it's possible to finish R&D 5 years sooner just because politicians have decided.  Now Shapps is saying it may be delivered to Ukraine even sooner than that because a system that is 70% done next year is better then one 99.9% done in 3 years.
More realistically, Ukraine needs any air defence it can get,  and the system gets to be tested heavily in real conditions, which will probably improve design iteration. So I guess we'll see whether it can become a meaningful and cost effective anti-drone system or whether its a white elephant.
Edit to add: whatever the rationale behind the decision making,  announcing it now has a lot more to do with timing of domestic and European politics, and the content of the announcement likewise.
https://www.bbc.co.uk/news/uk-68795603

New update from General Oleksandr Syrskyi:
https://t.me/osirskiy/650
 
 

155, GMLRS, ATACMS. Bradleys. All of these have been proven to work, they just need a ton more of them. NASSAMS and missiles for it. Then put real money and technical effort behind Ukraines drone programs. Last but most certainly not least  would be more training, preferably measured in months, not weeks.
 

If I may attempt to summarize, you have to win the war to earn the privilege of making the rules. And if rejecting a new technology means you lose....

If I may attempt to summarize, you have to win the war to earn the privilege of making the rules. And if rejecting a new technology means you lose....

World War II at sea was a race to understand a similar shift. In the span of a few years we went from measuring naval power projection in "weight of battle line broadside" to "size of modern air wing". Heavy surface ships survived inasmuch as they were able to be useful to the air wing projection assets, which was mediumly, and then not so much.
That transition happened in less than a decade.

I read the doc and RUSI actually hits some pretty salient points.  The mainstream thinking is that unmanned systems as we are talking about here are an addition to conventional warfare.  An emerging capability to be added to our extant capability portfolios and expenditures. Unmanned systems are an undeniably part of the future warfare military algorithms and focus should be on how to combine them best within our current approaches to create advantages.
I think this does not go far enough.  I believe that unmanned autonomous systems will emerge as the core pillar of a future military operational system.  We will then build the remaining systems, some legacy others also new, around these new unmanned capabilities.  We will fund and equip the unmanned forces first, along with C4ISR and PGM strike.  We will then need to figure out from the money left what to resources with respect to heavier conventional manned systems.  This takes the entire approach to force development and generation and flips it. More plainly, tanks will survive if they can demonstrate that they can shape, support and/or exploit the main unmanned battle…not the other way around as RUSI and others suggest.
This era we are in reminds me of the introduction of machine guns. Militaries of the day immediately brigaded them like cannons and relegated them to a support-to-infantry role.  The reality is that within a few short years the role of infantry was to protect the machine guns while they exerted firepower effect, and then the infantry would exploit that effect by taking and holding ground…so they could move up the machine guns.

This. The war in Ukraine is getting people thinking about it but the first terrorist attack on a city conducted by heat seeking, pattern recognition driven autonomous drones is going to galvanize the entire West.

We're all arguing this on a board that's dedicated to a wargame that has implemented at least some level of autonomy at the small unit level for 20 years.  And made it work in reasonable compute times for battalion sized swarms on computers that were nothing special.  The only thing it doesn't have is the physical sensor inputs, and those are pretty straightforward.  And it was all implemented by Charles and maybe a helper (I haven't kept up).  Charles himself might even count as an autonomous biocomputer, since he's really just a brain in a jar.

If I may attempt to summarize, you have to win the war to earn the privilege of making the rules. And if rejecting a new technology means you lose....

So just spitballing:
A platoon of M1A2 Sep V3 come in around $100M to just buy the platforms.  To this add logistical costs (fuels, parts and repair tail).  Force generation costs - training areas and exercises. Force projection costs (strat to op lift reqrs).  Pers costs - 16 crew, plus logistics, plus admin overhead.  And weird intangibles like route and bridge repair in training areas.
This all scales up very quickly.  A fully autonomous UAS swarm Bn may need a staff of 16 and a logistics tail but is not going to weigh roughly 300 tons that has to be transported and fueled.  Even at 1 million a pop, a hundred fully upgunned and militarized fully autonomous drones are starting to look pretty damned competitive against current “beyond-Night One” systems.

Uneconomical compared to what?

In the last several days there have been visits or major entreaties by Lord Cameron, PM Kishida and Pres. Zelensky. In addition there has been a pretty vocal group in Congress, both Democrats and Republicans, advocating for passing the aid bill. Their words were all publicized, at least for anyone who cared to see what they said. It's possible it's having some effect on public opinion.
Dave

I don't know if anyone has posted this yet, but  RUSI just published a paper on the present and near future state of drone warfare.
Mass Precision Strike: Designing UAV Complexes for Land Forces
by Justin Bronk and Jack Watling
Excerpt:
Swarming capabilities are commonly touted as the most significant area of capability development in the small UAV defence sector. However, the requirement to swarm introduces significant hardware and software complexity, which in turn drives cost growth and reduces the number of individual assets that can be fielded for any given budget. Massed UAV groupings, as seen regularly in light shows at civilian displays, rely on a ground control station tracking the position of all UAVs in a formation at all times and a central mission computer sending commands to each one to coordinate their movements. This allows large numbers of very simple small UAVs to fly in a coordinated fashion, but it is not a practical approach for military UAVs and weapons in a contested battlespace, due to terrain masking, EW, signal range and emissions control challenges – the ground control station would be struck, decapitating the whole swarm. Instead, for a mass precision strike complex to be capable of swarming tactics, the individual assets involved must have onboard sensors and low-latency datalinks that are resistant to hostile EW disruption. In addition, each asset must carry a mission computer powerful enough, and software complex enough, to fuse the information about terrain, threats and targets received from its own  sensors and those of other UAVs in the formation through datalinks, and to react to that information dynamically in real time. These capabilities are not inherently new, nor are they reliant on advances in AI or complex machine learning models. However, what the requirements for sensors, datalinks and advanced software do is raise component costs, even if used with an inherently cheap airframe/engine combination.
Furthermore, if a mass precision strike system is premised on swarming tactics for its effectiveness against its core target sets, then the number of assets required to use it in a sustained fashion will be increased, due to the need to consistently project sufficient assets into the target area to swarm. In conjunction with the increased hardware and software complexity required, this requirement to sustainably field swarming UAVs in large quantities over time means that fielding this sort of system as more than a ‘Night One’ theatre entry tool is likely to be uneconomical.
In terms of where swarming capabilities are likely to add value commensurate with the additional cost implied by their inclusion as part of a precision strike complex, the primary application will be to improve the capability to overwhelm air defence systems... Other advantages of swarming capabilities are that they can help reduce wasted warheads by deconflicting target selection so that multiple assets do not hit the same target. However, doing so in a way that can differentiate between a target having been hit and successfully disabled versus a target having been hit ineffectively and thus requiring a repeat strike with another asset requires significantly more advanced sensor and processing capabilities than simple deconfliction. Ultimately, for target deconfliction and strike optimisation, the value added question will come down to whether the additional efficiency against defended and undefended target sets gained from functional swarming capabilities outweighs the strike weight foregone by the increase in individual asset cost and the resultant reduction in quantity.

The issue with fully autonomous is that it offers superiority for a deterministic system.  That driver will pretty much ensure any attempts at regulation/proliferation are going to fall apart.  Now if autonomous systems achieve the level of a WMD with a MAD component, perhaps.  But the best counter to stop fully autonomous weapon systems...are other fully autonomous weapon systems.  We already have this in maritime warfare with missiles and point defence systems.  The CWIS is entirely autonomous once someone flips the switch.  They can target and engage on their own.  Why?  Because a machine can react far faster than a manned gun.
I don't think it is a question of Warhawk shrugging, it is the recognition that the odds of regulation that 1) we can agree upon and 2) sticks, is simply very unlikely.  Nuclear proliferation is a bad example because the morale imperative is not why the major powers did it.  They did so they could exclusively remain the major powers.  The other examples really are somewhat historical anomalies that we are also likely to walk back from as wars become more existential in nature.  Probably the best example is bio or chemical weapons, but we also know that neither of these really stuck either.
Trying to outlaw weapons is like trying to outlaw warfare.  We believe we can because we think that war is solely a political extremity and we can use political legality to control a political mechanism.  The reality is that the nature of warfare we currently subscribe to is the 2nd generation.  The 1st generation was "war is an extension of survival by violent means." That is the older darker nature of warfare that Clausewitz all tried to forget...right up to the point it throws itself in our faces.  In reality, we live in a third generation nature of warfare - "viable violence to achieve political ends."  The introduction of nuclear weapons put us all in a box whereby we can only really wage warfare in a constrained manner.  Go too far and one faces mutual annihilation.  The problem is when 3rd generation collides with the first one. 
So I fully believe in and adhere to the Law of Armed Conflict.  I think we should definitely aspire to be better than we really are.  But I know an existential capability when I see it. And fully autonomous weapon systems are definitely on that list.

I think many countries were probably quite happy to sign up to nuclear non-proliferation given the difficulty and expense involved in setting up your own nuclear arsenal.  Also, if anyone tried to breach non-proliferation treaties then there existed the legacy nuclear-armed powers who were able to carry out enforcement.
There will be basically no significant cost/difficulty barrier to establishing an autonomous killer drone fleet, once the technology exists.  That means any country will be able to do it almost at will, and, if they do, who would be able to stop them?  I think it'd have to be someone with an even bigger fleet, no?  Which means that, in this case, I don't think a treaty can work in the way we'd like it to.
Maybe the real answer is to stop thinking about developing multi-layered C-UAS as a way to free up areas to manoeuvre in southern Ukraine and to start considering it a matter of humanitarian necessity.

Saw this quote from Tim Snyder and thought it fitting to this discussion. It seems to me, Mike Johnson is looking ahead.

After this war....definitely a race to the bottom.  Ukraine had a $6B defence budget in 2021 - Russia had a $60B+ budget.  Ukraine's ability to stop Russia, push them back and now hold again is in no small part to employment of unmanned systems.
No government on earth is going to "just say no" on that level of disruption based on humanitarian or ethical reasons unless they are so secure that they can somehow take the high ground...looking at places like Iceland.  
What a lot of people in the AP mine and Cluster munitions camps did not understand is just how secondary or even tertiary these systems are to modern militaries.  Or maybe they did and were good with pushing the needle where they could.  Regardless, precision unmanned systems are by definition not indiscriminate. Will they be abused, most definitely.  But the opportunity/risks to the very legal and moral frameworks that would try to outlaw these systems is simply too high.  Fully autonomous systems are not a force multiplier, they are a deterministic weapon.  And as such, we are definitely headed towards them.

Now who is waving magic wands.  EW is definitely not infinite nor a perfect counter.  Pumping buckets of energy in all directions on the battlefield is the best way to get targeted and blowed up.  We have also seen plenty of HIMARSed EW platforms that prove this.  Also there are other ways for machines to communicate and synchronize than standard comms.  Image recognition algorithms, point to point low energy lasers and sound, to name a few.
The reality is that right now half the planet wants unmanned-power to be able to do what Russia is currently failing at - achieve superiority.  While the other half wants them to do what Ukraine is doing - create enough denial friction to stop a major power.  This means there will be enormous money thrown at this entire sector. 
And UGVs are just peeking out from the bushes.

I guess the economics for supplying and carrying an extra sensor package just for the last 20m are not there. Especially since radar works in that range, too.
Ultrasonic is only good if you need to detect stuff that radar doesn't bounce off - like a fly. So unless drones become THAT stealthy, sound is off the table.
 
Wrt to swarm communications: the swarm could communicate in ways that are designed to be low range. Like weak IR for example. No physical chance to pick that up from distance.
With mesh networks, the swarm can also be quite large if its density is high enough to relay the communication.
All doable - only question is when we will see it.

You mean like a strike package? With some flying CAP, others on EW, some clearing the route in, a couple providing oversight and a comms rebro and BDA, some SEAD, and of course some bearing warheads and payloads?
Yeah, of course. That all sounds clever and sensible, especially since it's already proven doctrinal approach to getting aerial effects delivery systems into an AO.
It doesn't sound simple or cheap though.

Yeah, that's why you might want to keep the communication rudimentary.  If the transmissions are short enough and infrequent enough, you can be somewhere else by the time most C-UAS systems are able to repond.  In a target rich environment, the comms would have to happen only very briefly and just before they all went in for their kills.  If there aren't a lot of targets, it wouldn't need to bother.