Comment by stcredzero | Hacker News Reader

Comment by stcredzero | Hacker News Reader

Parent Full thread

Have you ever shot a gun in real life

Yes. In JROTC.

developed an electromechanical system more complicated than an RC car?

Before arduino was even a thing, I wrote a bit-banging controller for a converting tiltrotor VTOL. (We were using the GPIO of a "Gumstix" micro linux board to produce the PWM signals for something like 6 motors with partial automation to continuously mix the vertical-mode and horizontal modes of control, dependent on wing tilt, so bit-banging a microcontroller was the only way we could do it at the time.) The model tiltrotor project was headed by a friend whose father worked on the actual LTV XC-142, so it was quite complex. The complete tiltrotor never flew, but the "Gumstix" did fly on a delta wing plane.

Because what you just described is a fantasy.

More like fodder for near-future spy thriller Sci-fi, but it is well grounded in reality.

The TrackingPoint rifles actually did work, and there is one with an 1800 yard range. The controller on a Linux system is the thing which actually pulled the trigger, relying on the operator to hold the rifle in the correct orientation. When it calculated there would be a hit, the computer pulled the trigger.

https://arstechnica.com/gadgets/2015/01/trackingpoint-shows-...

Mounting such a device on a large enough drone should be well within the capability of folks like CIA contractors.

> The TrackingPoint rifles actually did work, and there is one with an 1800 yard range.

TrackingPoint is attaching a very sophisticated electronics package to an otherwise standard rifle. The 1800yd range is based on a .338 Lapua round. That round requires a pretty hefty platform to remain stable.

> The controller on a Linux system is the thing which actually pulled the trigger, relying on the operator to hold the rifle in the correct orientation. When it calculated there would be a hit, the computer pulled the trigger.

There are two problems with this on a drone:

First, I don't see any information (understandably) about the allowable limits of muzzle motion for TrackingPoint. I suspect the system is tuned to correct the relatively small oscillations the muzzle goes through in the hands of a human shooter. I would love to see the drone platform that could hold a .338 rifle that steady half a mile in the air.

Second, the system calculates the solution at the point of the trigger pull. There is a brief period between trigger pull and the round exiting the barrel where barrel movement can throw the round off. This is fine if the movement is predictable (as it would be for a trained human). I suspect a drone is going to struggle mightily with a .338's recoil.

Drones can shoot down, drones can shoot up, drones can operate in free fall as easily as powered, they can operate in perigee and apogee, butter side up and butter side down.

Look at the ways the OP is (charitably) correct, picking holes doesn't make it impossible.

The 1800yd range is based on a .338 Lapua round. That round requires a pretty hefty platform to remain stable.

Yes, I was thinking of something much larger than your typical consumer drone. I wasn't thinking of the little Mavic drone that lands in your hand. I suspect that it would be much smaller than a Predator, however.

I would love to see the drone platform that could hold a .338 rifle that steady half a mile in the air.

I should think there'd be some kind of gimbal involved.

Second, the system calculates the solution at the point of the trigger pull.

If you mean, "around the time the operator pulls the trigger."

This is fine if the movement is predictable (as it would be for a trained human). I suspect a drone is going to struggle mightily with a .338's recoil.

You're telling me that CIA contractors or other engineers couldn't engineer some kind of gimbal that would be more stable than a human operator? Making it lightweight and incorporating it into a drone would be difficult. I don't think it would be impossible or impractical.

Since we are all cherry picking details (or taking pot shots might be a better phrase)...

A TAC-338 sniper rifle weighs in at a whopping 11lbs. There are off the shelf camera gimbals that can easily hold a 120lbs camera dead still from as far away as you care to film.

I don't think stabilization would be an issue at all for a drone mounted rifle.

> I should think there'd be some kind of gimbal involved.

Random thought: If the drone can take a moment to set up and snipe from range, what about dropping some ballast down a telescoping pole? It may not help with yaw, but it'd reduce tipping.

Shoot straight down, instead.

> This is fine if the movement is predictable (as it would be for a trained human). I suspect a drone is going to struggle mightily with a .338's recoil.

I was dismissing the OP at first, but you convinced me it's viable. This one detail (that is really the worst I could find) is the kind of thing where neural nets excel.

One would need to fire a few shots on training, but now I think precise targeting (in there being a small enough motor step) is harder to solve than this.

It only requires that to remain stable if you’d like to quickly fire another shot. Otherwise you can just let the rifle move: the round had left the barrel long ago by the time it moves even a tiny bit.

Reply on news.ycombinator.com