Hacker News Reader

Comment by lumost | Hacker News Reader

the pilot isn't the expensive kit, although the flight-hours to train them may be. Even an unmanned NGAD will probably cost ~500MM.

View on HN

I believe drones are actually projected to be smaller and MUCH cheaper.

We seem to be heading toward a plane with 1-2 people plus an AI that will cost $300M paired with 5-10 drones each costing around $30M. A real pilot nearby also reduces the efficiency of jamming to a large degree.

The idea is that smaller drones are harder to see on radar, they cost less, and they require many more missiles to fully intercept than one large drone.

People can withstand only certain G-s of acceleration. Remove them, and you have a much more manueverable craft.

Manoeuvrability probably isn’t as useful as it used to be. Fighter jets used to need to keep their radar, in the nose of the aircraft, pointed at their target to keep the target lit up for the missile to find. At close enough range if you can out turn your opponent and get out of its radar, you can “break radar lock-on”.

Today though, missiles have their own radar that works at close ranges but, if I’m right, still need guidance at long ranges. At close range, good luck out turning the much smaller faster missile. At long ranges you can’t really out turn anything.

Not to mention that with modern jets like the F-35, I heard the pilots can track you with their helmets, with the aircraft having sensors all over the airframe effectively allowing the pilot to “see-through” the aircraft, and transmit your location via data link to an in-flight missile.

If the ai plane can fly higher and faster than the rocket or at least a human plane because it does not have to care about a human body, then you have scenarios where the rocket must be shot significantly closer to catch the plane. Not much into military stuff, but seems logical that more options create more margins for success. Savings on inner space, human training, and equipment are only a bonus.

In most modern aircraft, the aircraft over g's long before the human body. When aircraft are pushed to the limit in a combat situation, you'll hear the warning alarm blaring "over-g, over-g" as continued operation in that domain will either tear the wings off - or permanently degrade the airframe.

The main advantage of a drone is expendability - sending a pilot to certain death is not something the military wants to do. However there are missions where risking a 300 MM drone on a 30% chance of destruction makes sense. A missile truck drone with an AESA radar could likely maintain a 4:1 kill ratio by simply flying further into enemy air defenses than a human pilot would. It doesn't matter if you lose the drone if it kills 4 other air defense targets worth 300MM.

Sure, but you base the assumption on the ai being deployed on a current design, and not on something developed without that constraint. I'm not sure if such designs are viable, but someone is sure to look into it.

aye - my assertion is that the limits of the human body are not a major limiter in current military aviation. You can see an example of how current military planners are thinking about things with the B-1R concept (ultimately rejected, due to overlap with F-15EX). The limiting factors are observability, radar, sensors/data-link, payload size, range, and speed. None of these restrictions are problematic for a human pilot.

Maneuverability improvements such as super-maneuverability were rejected due to being counter-productive in Beyond Visual Range combat. You have to give up to much energy for fancy acrobatics to be worthwhile.

The AIM-260 can be guided by AWACS to within ~5 km of target, where it then turns on it's own radar for final targeting. Generally, the expectation of a modern F-35 pilot is to act as a missile truck with the basic A2A/SEAD/CAP mission being to fly within a viable area to deploy the missile, release, get out of dodge. Of course, there are hundreds of counter-measures aimed at disrupting this mission, and counters of counters which the pilot must be an expert in.

Nope. In order to make the aircraft more maneuverable you also need a stronger (heavier) airframe, different inlet geometry (to avoid compressor stall), more powerful (less efficient) engines, etc. For a multirole aircraft those are bad tradeoffs to make. And removing the human pilot costs a lot of tactical flexibility: autonomous flight software and sensors can still only handle very limited missions, and remote piloting links are unreliable.

Dogfights are not a thing anymore. A2A missiles no longer have a 30 mile dud rate and if you want to close to gun or infrared missile range, you have 50-100 miles of eating AMRAAMs or even longer if you are facing a Russian plane. Air to Air combat hasn't been constrained by human G force tolerance since late Vietnam war.

The flight hours + HR costs.

The key financial advantage of drones is that they don't need money to live after work, health insurance, or a pension plan. It's expensive to ask people to risk their lives for work.

And specifically for forward deployment scenarios: they don't require human support infrastructure pushed forward, at great cost.

Even in austere military terms, it's logistically expensive to keep a human alive on a random atoll in the Pacific.

The aircraft:supportHumans ratio is a tyranny in that each of those supportHumans requires their own support.

>It's expensive to ask people to risk their lives for work.

Unless you're in the Russian army.

The Russians are paying good-for-Russia bonuses to attract recruits, as well as giving anyone who fights or builds for the military in Ukraine veteran status.

They're promising good bonuses for signing up. There's a lot of complaints about people not actually getting what they've been promised.

But will you survive to enjoy those beenfits?

Also you can build a drone AI and cheaply deploy it to 1000 drones. Whereas each human pilot must be expensively trained.

Reply on news.ycombinator.com