2021/09/18

Uniforms are an obsolete concept

.

Everybody nowadays agrees that the red, blue, white uniforms of the absolutism era and even 19th century would be a horrible idea on a modern battlefield. They made troops especially visible and identifiable, which made the commanding of formations on a battlefield easier. Firearms improved in firepower, and it became imperative to seek concealment, cover, camouflage and be dispersed.

The French persisted in using colourful uniforms into WWI
(correctly colourised photo)

The reason why such easily visible uniforms are obsolete is that they're too easy to detect and identify. You cannot afford that on a modern battlefield because firepower has become terribly lethal (and the partial bulletproofing of infantry doesn't change this).

That's EXACTLY the reason why I think that ALL uniforms - including modern ones - are obsolete.

What? You think it's not that easy to see a camouflage pattern-clothed soldier outdoors at 100 m?

Me neither, but you should stop thinking with your eyes. This is the age of electronics. It's a decades-old story that law enforcement uses aerial sensors to find marijuana plants based on their colour spectrum.* This is literally something that can be done with a man-portable drone over several kilometres range.

Likewise, all those exactly military specification-following camouflage clothes and vehicle camouflage paint jobs (and even the industrially-made ghillies and camouflage nets) can (as far as I know) be detected quite easily by their spectral fingerprint. They stand out to a appropriately-prepared sensor and computer as bright red would do.

A typical man is unable to correctly name more than two or three shades of green. Some Amazonian tribe reputedly knows 50 names for different shades of green. A computer can correctly identify ten thousands of shades of green in less than 1/1000th of a second.

Weathering of colours (such as many times washed clothes, or old vehicle paint jobs) is very likely something that the algorithm can simply take into account. The false alarm rate may be an issue, but today's machine learning methods and an ability to zoom in on a first detection for confirmation should render false alarm rates acceptable.

Camouflage patterns are furthermore patterns. Repeat, patterns. Pattern recognition is one of the strengths developed for computers in the past two decades. Camouflages patterns are difficult to discern from certain backgrounds for a human brain at certain distances - but a dedicated computer algorithm can identify such patterns easily regardless of context.

 

In the end, technology may spell doom for uniforms and standardised vehicle paint jobs for battlefield usage, leaving us with uniforms for non-battlefield troops, which would be a relic - largely devoid of function.

The things that may be redeemable about battlefield uniforms are their shapes (cut) and materials.


Predictability hurts in warfare - and what's more predictable than an army wearing the same clothes and painting all its vehicles alike, for decades?


edit: a somewhat related research paper

https://www.scientific.net/AMM.135-136.341

Figure 4 shows the problem; the amouflage fabric looks like foliage to us humans, but a spectrum analysis shows something completely different. Figure 6 reminds me of how algorithms can easily discern a small boat from a background of seawater and white spray - the effective contrast is much greater to the algorithm than to the human brain.

S O

defence_and_freedom@gmx.de

 example: channelsystems.ca/sites/default/files/documents/UseofSpectralPropertiesforWeedDetectionandIdentification_Noble.pdf  If still in doubt, look up 'marijuana spectrum detection', but absolutely avoid searching for 'marijuana spectrum'...

P.S.:  Years ago I wrote that camo pattern uniforms are fine for non-combat, non-reconnaissance troops, but combat and reconnaissance troops should strive for better camouflage effect than possible with standard pattern uniforms. Ghillie-like camo jackets modified for the local and seasonal environment, for example. Camo pattern clothing risks suggesting that it's good-enough camo by itself, but both combat and recce troops should strive for better stealth than that. One way to ensure this could be to give them monochrome grey, green or brown camo clothes (at least for the torso). Now I think that maybe the visible parts of such a colour base layer could be very troublesome.

.

19 comments:

  1. Pattern recognition is both the biggest strength and weakness of computers: they will see things that aren't there, just like we do. But they lack intuition and other means to filter the noise(as we lack computer power and emotional detachment, but there are inherent limitations to everything). The battlefield can be flooded with "noise", cheaply made, dumb drones, which carry the camouflage pattern, and move in swarms, following military doctrine. They could be the recce/light attack attached elements of the army. Camouflage pattern could develop into something more akin to dazzle paint, hiding the shapes and orientation of the object, and thus helping to fill with false contacts and noise the enemy's sensors. More than concealment, the new uniforms would be made to confuse the sensors into the true nature of the contact, by messing with the pattern recognition software: is it a drone? A man? A tank or a truck? Also, malware and viruses can be uploaded via pixel patterns, adding another layer to the ECM defenses.
    Patterns can be made even harder to detect, by increasing the repetition interval, or using other areas of the electromagnetic spectrum in the color combinations. Maybe the pattern only starts repeating every 15km of fabric, and it assumes sensibility to 8-12 colors, instead of three and IR/UV in general. That forces a lot more computing power to be used, which means the drones would have to use cloud computing tech to process the deluge of data. That would make them more vulnerable to jamming and interference, and may help triangulation of the enemy forces.
    I do think that more low level customization and modification is in order. The camouflage must be more like the skin of wild animals, with fur in other, patterns of color. The uniforms would then have to be always modified by the combat troops with a "fur", whose value is on the materials and randomness employed.
    The elefant in the room might be that the brass and politicians may have forgotten that the ultimate virtue of military dress is to be comfortable, protection from the elements, light, and cheap to manufacture in the millions with an reasonable quality. Soldiers will spend a huge amount of time doing many, many other tasks than combat (as you have mentioned in several posts). The Israeli way, of having a uniform meant for work and wear, and add-on camouflage may be the way to go, and would add more noise, by having panels cut from different parts of the fabric roll, adding randomness to the uniform.
    The alternative would be using... Civilian clothes? Quality is abysmal in 90% of the times, and their price depends on extremely long and fragile supply chains. Protection from the elements and comfort are debatable for a huge majority. Civilian outdoor clothing may be better, but it is expensive as hell, and designed to be easy to spot in case of emergencies. And, again, made in far away countries, with fragile supply chains, and materials that are not that easily made locally.
    A weird and illustrative example of this kind of material is carbon fiber. There isn't a single carbon fiber factory in Europe. None. There are plans to build one in Portugal, with demand under severe stress due to coronavirus measures, but it could take years to get up to speed. Gore Tex and other advanced materials are in a similar situation: the clothes may be easily made in Italy, Croatia or elsewhere, but the fabric and threads are made in East and South East Asia. They are still a labor intensive process which takes years to reach an acceptable level of quality.
    The marijuana drone detection program is a good example of this noise-approach: in spite of the computer prowess, it is just easier to spread the crop and to put many, many smallish fields all over the countryside than to create huge ones. It is estimated than less than 10% is ever detected and/or destroyed, and the supply has grown steadily, with prices relatively low, over the years...

    ReplyDelete
    Replies
    1. The carbon fibre thing is completely wrong.
      https://www.statista.com/statistics/380553/distribution-of-carbon-fiber-production-capacity-worldwide-by-country/

      Delete
    2. https://www.nrel.gov/docs/fy17osti/66875.pdf

      page v

      Delete
    3. "A weird and illustrative example of this kind of material is carbon fiber. There isn't a single carbon fiber factory in Europe. None. "

      That sounds for me as chemist wrong, very wrong. Central Europe is home of a lot of high tech fiber production. Carbon fiber is not that special.

      Delete
  2. If the fabric reflects light in a common spectrum and in a common natural pattern, it won't be easy to detect. Current designs have the limitations of human eyes in mind that can't dfferentate many shades. One example would be gluing earth on it to look like earth.

    ReplyDelete
    Replies
    1. Keep in mind marijuana leaves look common green, and can easily be discerned by aerial sensors.

      Delete
  3. One thing about camo patterns today is, that most camo patterns do not work the same for aerial observation than the do in ground observation. Many great camo patterns that work on the ground well fail against aerial observation as i have tested several times with a uav. It was stunning how different it was from above.

    This would be an true advantage for such drones against the conventional actual camo patterns.

    I want to add one thought: if camo patters and uniforms wil become an disadvantage the only logical solution is to hide the soliders amongst the civilian population in civilan clothes.

    ReplyDelete
    Replies
    1. This comment has been removed by the author.

      Delete
    2. @pootis
      Feel free to provide a source. I read the GC and don't remember such a thing.
      It's a war crime to keep civilians from moving away from soldiers in wartime.

      Delete
    3. @S O

      "Article 44(3) of the 1977 Additional Protocol I provides:
      In order to promote the protection of the civilian population from the effects of hostilities, combatants are obliged to distinguish themselves from the civilian population while they are engaged in an attack or in a military operation preparatory to an attack"
      found here: https://ihl-databases.icrc.org/customary-ihl/eng/docs/v2_rul_rule106_sectiona
      It is not allowed for soldiers to hide among civilians

      Delete
    4. No, that part only says they have to mark themselves. A brassard with an official symbol of your armed service or country suffices for this.

      You may still run from a sensor drone into a Kindergarten and hide there as a soldier. What you are not allowed to do is to keep the civilian staff and the children from moving away from the vicinity of legitimate targets (such as you, the soldier).

      Delete
    5. i was originally responding to the 2nd part of Anonymous's comment. Who suggested to hide soldiers among civilians

      Delete
  4. It's always seemed to me that camouflage patterns, like uniforms in general, serve as much a function of identity as of concealment.

    ReplyDelete
    Replies
    1. The Western allies supposedly refrained from using camo pattern clothing in Western Europe 1944/45 because of fear their troops could be mistaken for German troops who used some patterns (the Americans used patterned camo clothes in the Pacific theatre).
      We have long since given up such notions.

      Delete
  5. This is a great post.

    "My reflection, when I first made myself master of the central idea of the ‘Origin’ was, ‘How extremely stupid not to have thought of that!’ " Huxley, Thomas Henry. 1887.

    Modern production lines strive to reduce variation, since some variation will be unaceptably low quality.

    ReplyDelete
  6. I think that a part of what keeps the current paradigm, so to speak, going is that the thermal imaging stuff is still very rare, very expensive, and in very few hands. They are still reliant upon people using unaided vision, or at best binoculars that would have been unremarkable a century ago. This is very much like the case of the satellites in orbit that can supposedly read license plate numbers. There are very few of them, they have very, very narrow fields of vision, and the world is very, very big. Without someone on the ground who can see something interesting without it, then get on the radio and say "Have a look at what I found at these coordinates," they're much less useful and powerful than the sales brochure implied. So the camouflage uniforms and camouflage painted vehicles and fortifications remain as they are.

    ReplyDelete
  7. Modern sensor and data processing does improve detection and identification, but the effective range is still not infinite. Designed materials still reduce the range. Randomized patterns maybe useful for increasing effectiveness, however I'd expect the final result to fall into a small design space relative to "all clothing" and a combatant still have a specific look in general, even if no two piece of clothing is the same. Any nation with an AI industry can reverse search for optimal designs.

    The addition of algorithms to the picture means effective camo can involve intelligence work: either reverse engineering opponent sensor algorithms or capture of camo patterns can influence subsequent performance as vulnerability are discovered or patched.

    One can expect behavior like code breaking success of the past where information is willingly kept from line units for fear of leaking it to the opponent. If patchable algorithmic vulnerabilities are found (see adversarial images) it would only be implemented to a small subset of combatants for high value operations or ones that do not risk leakage of information.
    ----
    In the inverse, clothing (not uniform since it may be different with regard to individuals) with visual cryptographic identification functions built in may be used in low intensity conflicts. Without it a persistent deep fakes campaign can really muddle the information environment.

    ReplyDelete
  8. I think fabrics that will change colour and camo pattern will be available soon. This must not be hyperstealth magic stuff, for example the czech army produced fabrics recently which change colour with temperature and/or with small amounts of chemicals distributed other them. This together with an kind of half-ghillies (upper body only, also useful against thermal vision) is imo the future for infantry.

    This inevitable creates the need to sacrifice body armour and helmets, as this add heavily to thermal signature and are also to much an hindrance for the continous movement necessary in modern warfare.

    So the experiences from the occpution wars of the recent decades lead to an complete wrong direction for the infantry, especially in the value of stealth.

    ReplyDelete