[CoolGuySteve]

One of the reasons the attempted drone assassination of Maduro in 2018 failed was because police were using radio jammers: https://en.wikipedia.org/wiki/2018_Caracas_drone_attack

It's fairly easy and cheap to make an explosive drone and train a neural network to drive it using off the shelf components and open source software. It follows that it's inevitable that at least semi-autonomous drones will be made to counter radio jamming.

So while the best outcome a treaty could hope for is to prevent mass manufacturing these things by some countries, any motivated small country/individual will be able to craft them.

[gonzo41]

Drones can't beat nylon nets or other low tech countermeasures. I think anti drone water cannons will outpace drone assassinations. And more than both, regular bombs and guns will still do most of the killing.

[gpm]

A motorized sniper rifle with face detection to find targets and a high rate of fire seems like it could exist with current technology, require about the same effort to deploy as a sniper with a typical sniper rifle, require less training from the person in the field, and be a hell of a lot more deadly. Same with a machine gun for closer quarters.

Regular bombs and guns probably still do most of the killing for quite awhile because of inertia, but I'm pretty firmly of the belief that they are outdated technology or will be soon, especially the regular guns part.

Regular bombs are arguably already outdated by missiles with (frequently autonomous) guidance... but for a lot of applications dumb explosives do work pretty well so those probably stick around for a long time.

[nonameiguess]

While it isn't impossible in principle to make an aerial stabilized firing platform, I think you underestimate the difficulty. You definitely can't do it with a quadcopter, which doesn't have anywhere near the precision reaction capability to counteract rifle recoil. Even the most advanced fighter jets in existence right now can't do this. They need to use guide missiles for precision targeting.

In practice, a drone swarm is an area weapon. It's why it's kamikaze and relies on explosives right now. It can say it has the ability to find a specific person using computer vision and maybe it can, but it definitely can't make sure it only kills that one person and it can't kill from a distance.

No flying platform will be replacing snipers any time soon.

[laverya]

> You definitely can't do it with a quadcopter, which doesn't have anywhere near the precision reaction capability to counteract rifle recoil.

I think you're overstating the difficulty here. There's three things a drone needs to do:

1. Provide a stable/stabilized platform (for sensors and effectors)

2. Respond to recoil in such a way to make the point-of-impact consistent

3. Not fall out of the sky in response to recoil

I think 1 is demonstrated today with drones flying around 4k/8k movie cameras - a lot of that might be the mount, but you could put an AR-15 in that mount too. 2 is presumably easy as well - just make sure that the drone is in the same physical orientation before each shot as in 'training'. 3 is trivially solved - if your drone is robust enough to withstand the recoil, just let the computer autolevel the drone in whatever new location it's found itself in. If this costs altitude, make sure to fire from farther up.

[bombcar]

Do you even need to counter recoil if you only want to get off one shot?

[leetcrew]

newton's third law says yes, at least to some extent. unless the barrel is perfectly aligned with the platform's center of gravity, it will be deflected (at least a little) before the projectile exits. and it's not trivial to hold a quadcopter or fixed wing craft steady enough to aim precisely in the first place.

[jhgb]

Do you need to aim steady or do you only need to fire precisely at the right moment? For example, TOWs using a top-attack EFP warhead don't "aim steady" either, they just detonate at the right moment.

[gpm]

I'm not suggesting a flying platform... but a ground mounted one moved into place by a human.

[prussian]

Hell you don't even need to go that high tech. Just make model jet aircraft cheaper, easier to handle, maybe even autonomous and I'd like to see those current anti-drone countermeasures stop it from zooming in and dropping whatever dumb payload someone could possibly want.

[gpm]

Honestly I feel like a model jet aircraft is a lot higher tech than my proposal :P

But certainly lots of ways to make use of autonomy, and I'm no expert on the best way.

[gonzo41]

Sniping isn't that simple. The risk of counterfire is real. I think autonomous weapons will always be more successful in area denial and holding territory. Ie landmines.

[gpm]

The risk of counterfire is a lot smaller when the only thing to fire on is the few thousands of dollars worth of gun and electronics... and you could even armor it pretty effectively at the cost of mobility.

[namlem]

It's not that easy. A startup tried to build a smart gun with aim assist, but couldn't keep it calibrated for more than a couple shots and eventually went under.

[jhgb]

Can't one drone destroy your nylon net for the rest of the drones to fly through it?