1
$\begingroup$

Imagine something like the M2 Bradley, except with no passenger capabilities and a single crew member.

I recognize that having one crew member is rather harmful redundancy-wise, but that's not relevant to the scope of the question - I want an AFV with one crew member for other reasons.

Advanced fire control systems assist with aiming, five machine learning algorithms can make decisions on how to get from point A to point B (three need to agree in order for it to take a course of action), and there's an autoloader - among various other labor/time-saving systems - but the crew member decides on where Point B is, and makes all actual decisions related to anything from firing the weapons systems to radio communication to where to drive to.

Basically, the crew member tells the systems what to do, and they execute on it. That part's already been hashed out; it was the reason that one-crew tanks were previously impossible, because it put a lot of work on one person. Now, computers can do the work.

What would such an AFV look like? For instance:

  • Would the crew member sit in the turret or hull?

  • Where would the turret be positioned? It's not like there's much crew compartment to get in the way.

  • With modern limitations in computing, would the computers/automation support systems be more volume-efficient than a human?

  • Would this make the vehicle cheaper in the long run, since it costs less to CBRN-proof it, and paying one crew member's salary is less expensive than, say, that of three?

  • and other such questions

$\endgroup$
6
  • 1
    $\begingroup$ If you're using AI mediation to turn the high-level crew member decisions into specific actions, why do they need to be present? Why not a drone? $\endgroup$
    – Cadence
    Sep 1 '21 at 18:09
  • $\begingroup$ @Cadence Crew members are capable of performing maintenance more effectively, cannot be effected by electronic countermeasures, and provide "man-in-the-loop" capability. Modern-tech AIs might make very dangerous/war-crimey decisions, and are vulnerable to errors (for instance, classifying a dog as a cat), which is why there are 5 of the things in charge of navigation rather than one. They're there to carry out repetitive tasks that would otherwise eat up the crew member's time/energy. $\endgroup$
    – KEY_ABRADE
    Sep 1 '21 at 18:56
  • $\begingroup$ What would be the tactical role of this AFV? I presume it should be a main battle tank, but I see that you are staring from M2 Bradley example. $\endgroup$
    – Alexander
    Sep 1 '21 at 20:01
  • $\begingroup$ @Alexander Suppression of rebel/insurgent forces in heavily mountainous terrain. $\endgroup$
    – KEY_ABRADE
    Sep 1 '21 at 20:56
  • $\begingroup$ @KEY_ABRADE I see. Do you want a traditional tracked/wheeled machine, or we should improvise and make it a mecha? $\endgroup$
    – Alexander
    Sep 1 '21 at 21:34
4
$\begingroup$

One of the main issues here would be task overload. Even with AI taking over many of the aspects, it is difficult to simultaneously (1) plot a safe route in a combat environment; (2) observe the surroundings for signs of enemies; (3) maintain proper communication and coordination; and (4) correctly use the weapons available to you.

While video games like Battlefield, Call of Duty or Halo have you control vehicles like AFVs or tanks as a single person, a key difference is that very often the enemy has been designed in these games to be visible and easy to identify (with indicators over their heads if not by visual design like in Halo). In any realistic scenario, this is not the case. Any half-intelligent enemy will be trying to conceal themselves as hard as they can until it's too late for you.

So, to help your one-person AFV:

  1. They should sit in the hull, low down, preferably behind an engine block or a nice big hunk of armour so they're less likely to be killed if hit

  2. The turret should be positioned wherever it has best access to seeing and shooting things. Most likely in the centre of the hull. It should be as thin as possible to make it difficult to shoot (see Russia's T14 Armata - the actual turret is actually not much wider than the gun base, but some extra armour and other sensors etc on the sides makes it look much wider).

  3. (and this is the important one) They absolutly must have access to a comprehensive suite of top-notch sensors, including wide-angle visible spectrum, IR and potentially others like high frequency Radar for low-visibility situations. These should be helped by a suite of augmented reality features run by computers to assist in tasks such as automated target identification, threat location (IEDs, explosives, etc), battlefield awareness (Friend or Foe). They should also be linked through intuitive displays, or else the operators should undergo extensive training.

For these vehicles to be effective, you essentially want to offload almost all target acquisition, weapons handling (aiming and firing accuratley), battlefield awareness and driving to automated systems. You want to almost turn the battlefield into a video game for the operator, where the operator of these vehicles acts purely as a decision-maker - their role should be equivalent to that of a tank commander, where they are responsible only for locating targets, deciding what to shoot at, when to shoot at it, and coordinating/communicating with friendly assets.

I suspect that unless there were massive advancements in computing power, size and cooling (which would be massive considerations for these vehicles), then having all of this very expensive technology and systems on these vehicles would make them more expensive to produce and maintain than systems with more crew. Think of the difference in cost between an F35 and an F16 fighter jet, for instance. It's not just the stealth aspects that costs money, but the advanced sensors, computing and other systems (and all that maintenance!).

Perhaps you might want to look into the lore of Battletech for some inspiration: it has single operators working large mechs, and it's often emphasized how there is a dizzying variety of sensors and systems that the operators have to proficient with, as well as years-long training required to reach even basic proficiency.

$\endgroup$
3
$\begingroup$

Pro forma crew member.

/(5) Certify that operators of autonomous and semi-autonomous weapon systems have been trained in system capabilities, doctrine, and TTPs in order to exercise appropriate levels of human judgment in the use of force and employ systems with appropriate care and in accordance with the law of war, applicable treaties, weapon system safety rules, and applicable ROE./

Department of Defense Directive, 11/12/2012. Autonomy in Weapons Systems

After some terrible events it was mandated that autonomous weapons have a human operator ready to take control at all times. Because of the ease with which remote human operation can be disrupted, the human operator must be physically present in any autonomous weapons system.

The current setup honors the letter of the law if not the intent. A human crew member is present within each autonomous weapons system including these AFVs. The crew member is not well trained or particularly skilled, and chosen primarily for his or her willingness to ride around inside a killer robot and exclaim "Yes!" when asked questions by the robot. The pay is good.

--

It sounds silly and that would be the initial impression. But what if one of these half-sober physically unfit crew members realized some very bad stuff was going down outside the autonomous robot. Can they take control? What happens when they do?

$\endgroup$
1
$\begingroup$

Aight, first things first.

You want Anti-insurgency, right?

It is likely to use guerilla tactics, which means hiding. Your vehicle does not have human eyes scanning the ENTIRE FOV so you need to automate that.

I suggest a false canopy system within an enclosed cockpit, becaude theres only 1 pilot, so you need to protect that(and plus your automating so cost is likely no issue)

depending on what environment your in(desert, mountain forest, sno, etc) you likely want at least a thermal imager(as well as a normal camera, thermals ain't always best on) in every direction you expect an enemy.

A large sandy field yields an advantage to who has the longer range guns. A hand carried atgm will likely not outrange the main gun of the tank you fire it at.

In an urban envoronment where you can hide in a building, you want the ability to look up. If you have it as a mecha thats tall enough you also need to look down.

Same with forests, you also can hide in a tree OR in a bush.

So yeah sensors EVERYWHERE.

Also, do you want legs, tracks, or wheels? Personally if you can (technologically/if you want) I suggest a robust mounting system that allows optional mounting of each(like a hand that can hold many different tools). A "waist" that mimics a turet ring of many tanks that can mount

  1. Some hips to attach legs
  2. A chassis that has wheels
  3. A chassis that has wheels

Also, I suggest some way to manipulate the outside environment, as say I single obstacle can be real annoying when you have all this destructive force and yet this rock or tree won't let you through. So slap an arm or 2 on it.

Cooling. All those automatic systems WILL generate LOTS of heat. Pilots won't like that, at all. Get the best cooling system you can, you'll need it.

Also I suggest having a few different weapons on hand. Sure, your .50 can shred human, but that concrete block they are hiding behind does't care, it still ain't moving.

Here, something like lobbing a grenade behind would be nice(I suggest automating the aiming of all these weapons, the pilot has more things to do).

A nice set of automatic machine guns(say, a .50 bmg or am mg3) and some explosive weapon(say, a large cracker grenade in which ghe fragments ARE your standard infantry grenade, or at least the same explosive part)

Overall, be ready for lots of different things, be able to see anywhere, and overall make operating said vehicle feel similar to how video games have taken to a single player operating a multicrew vehicle.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .