If the goal is to hit a target, the robot can do something a human cannot: fire multiple shots quickly. The camera would tell it exactly how far off the first shot was. A servo can make micrometer adjustments and fire again before the target can react or move. Presumably, wind doesn’t usually change much within such a small time window.
I find this difficult to believe. It seems to me that a properly designed cruise missile would have redundant fusing and navigation systems.
If you’re interested in the Hollywood version, there’s one such rig used in the movie The Jackal (with Richard Gere and Bruce Willis). Although to be fair it’s not 100% robotic, more of a remote controlled heavy machine gun.
Believe it. The Phalanx CIWS fires 20 mm caliber autocannon rounds with a muzzle velocity of 1,100 m/sec. See the picture in the attached link. The 20 mm round is the one that is bigger than the 0.50 caliber BMG rounds in the same picture. Here is a picture of the M61 Vulcan cannon used in the Phalanx CIWS.
One of these rounds will do quite a job on anything electronic, especially in a head-on shot, which will penetrate through the nose cone of the missile like a hot knife through butter.
Anti-ship missiles travel very fast, and this is their primary defense. They are not designed to withstand much battle damage. They are designed to not get hit. If they do get hit, all bets are off.
Missiles are expensive enough without adding redundant systems. Military designers usually save redundant systems for manned aircraft and vessels.
And yet human snipers still manage to hit targets. They aren’t doing anything magical; firing bullets is purely kinematics. If a human can make a compute a good firing solution, then so can a computer, given the same inputs.
This is not yet the case, and may not be for a very long time. Humans have a kind of imput it’s not possible to practically give the computer in a very short length of time.
To account solely for wind alone (and nothing else, an extremely generous situation already), we have to assume a robot capable of measuring wind on its position constantly and accurately (doable but likely bulky and computer-hardwarre-expensive). Then it would have to take into account all visual information downfield to calculate wind at various points there, as well as wind at the target. Plus, take into account various pressure conditions for an accurate reading, etc. And that’s one variable.
So, no, we can’t do it now, because while computers can make a good shot calculation, if given all the information, that’s not practical for the foreseeable future.
And what’s the gain from doing this? You’d then have to still get people to carry it into the field, it would still have to be tough enough to endure field conditions, it would need highly developed servos on a very small scale, and it would still require people in the field to spot for it
I think a robot may not need to account for wind because a simple difference engine requires only three shots: Fire first shot, note where bullet lands relative to bullseye. Fire second shot with an adjustment towards bullseye, note where bullet lands relative to bullseye. This gives enough information for a final adjustment. The third shot should pretty much always nail the bullseye.
If the sniper is far away, the target may not even hear the shots.
As for what’s to gain, it’s like a cameraman undergoing a lot of time-consuming training to hold a camera extremely steady and learning how to press the button slowly vs. anyone throwing down a tripod and using that cable presser thing to snap the shot. Both result in a crystal sharp picture, but one person is using better tools.
If the average soldier doesn’t have the eagle eyes, steady heartbeat / breath control, and sniper experience and wisdom to take a far away shot, they can still bolt down or anchor a tripod, press the “zoom” button on the camera, click on the target, and accomplish the objective.
I guess if your target is standing against a nice bright background so the camera can identify the bullet holes, this might work. How would it handle it if the target were in the open, and you missed by an inch or two…I don’t know of many cameras small enough to fit on a rifle that would have the optics and processing power to actually track the round in-flight like you are saying.
Can we get a cite that human snipers can consistently hit a human target on the first shot, from any range where wind is relevant? Yes, it certainly happens, and we can all find anecdotes of someone or other doing it, but can they do it consistently? Because I suspect that the only practical method for tracking wind conditions everywhere between the shooter and the target is to fire a shot and see how much it missed by. And it’s not like “three shots, one kill” is practically any worse than “one shot, one kill”, other than for purposes of bragging rights.
DARPA were sniffing about for a rifle scope that would do that a couple of years ago. No idea if they have got anywhere - downrange laser crosswind sensors were hot news [PDF alert] for a while but seems to have gone a bit quiet for the last few years. Presumably it proved harder than anticipated to sneak up to the target and erect a nice shiny laser reflector.
crazyjoe - that particular problem could probably be solved by using ammunition with flash compound in it to give a nice distinct burst of light (of fairly distinct spectrum) from the point of impact. Or use IIR to track the projectile in flight. But all the optics and processing necessary certainly tend to push towards an AFV-turret rather than a rifle-sight solution.
The Controvert - I think that there’s also a disconnect between what’s technically possible and what’s necessary. The DARPA boffins could probably knock up something in a few months that do exactly what you need. However it would be huge, hideously expensive, and would be limited by the ballistic capabilites of rifle ammo. And if you’re that desperate to kill someone, there are far better alternatives already issued at squad level or available on the end of a radio.
Why bother building a $5mm robo-sniper to kill someone when you can just blow them apart with an ATGW, laser-guided mortar round, JDAM, an AFV cannon shell or by just squirting a couple of hundred rounds of normal ammo in their general direction? Crude and messy, but usually good enough for government work.
BUllets for sniping are supersonic. The target doesn’t hear the bullet, because he’s been killed. Everyone else around for a mile will.
Except you still need expert scouts with most of the same training. Sniper training is not mostly marksmanship. That’s the easy part. The hard part is learning to get in and out alive.
Look, to be blunt, the entire pro-robot side here is ignorant of the demands in the field, ignorant of the technology required, and ignorant of the concet of “cost-effectiveness.” If you’re so bloody certain it will work, go make it. Then you can come back and complain we weren’t fair to you. Until that day… no.
The Discovery Channel has had several shows about snipers. In one, a sniper team put all 7 shots into a man size target from 1000 yards. This was done on a rifle range so distance was known but there was a wind present. Two ranging shots were fired, both hitting the target, the next 5 were as fast as the shooter could work the bolt.
My understanding is that distance will pretty much always be known for a long-distance shot, either because one of the team has ranged the target or because while they were waiting for the target to show up they ranged every landmark within sight and wrote it all down.
Having recently read a book written by a British sniper, taking super-accurate long-distance shots at a precise target is only a tiny bit of what their job may involve. Figuring out where shots are likely coming from, working out which bush or bit of junk the target is hiding behind, waiting for the target to be legitimate under ROE, diving into cover to avoid incoming mortars, hiding clingfilm-wrapped turds in each other’s rucksacks, sneaking about on people’s rooftops to set ambushes, snap-shooting fleeting targets, it is indeed a very varied job.
At long ranges, bullet drop from, say 900-1000 yards is considerable so even a very small error in ranging can mean a miss. Being on a range where you know 1000 yds. is exactly 1000 yds. takes out one variable.
Which is evidence that sometimes they get lucky. If you hit with your first shot, then all you have to do is keep everything you control the same, and fire fast enough that the things you don’t control can’t change much, and your subsequent shots will hit, too. What I would like to see is for a sniper to be put in some situation, take a shot, then be put in a different situation, take another shot, and so on, and see how many times he makes the kill on the first shot. And, of course, record the results of all shots, not just the ones that look most impressive on TV.
Does anyone know the magnitude of the effect of a consistent side wind on a bullet fired from a typical sniper rifle at 1,000 yards?
For example, if wind was 5mph vs 10mph vs 20mph, etc.
Are we talking a couple inches, or is it more substantial?
At say 1000 yards, what is more acurate.
A .50 Barrette?
Or a 120mm with a sabot round?
(I’m not suggesting that a tank could sneak up and perform the same task as a sniper)
Actually, recoil always guarrantees the situation is different (greatly so) in rapidly repeated shots.
Then you put the gun back where it was before the recoil moved it. If I aim for a spot 3 feet left of my target and 18 inches up, and my first shot hits, then I’m going to aim my second shot 3 feet left and 18 inches up, too. I assume that the sniper is still looking through the scope and controlling the rifle as he makes the subsequent shots, no?
Indeed. Call it sixth sense, but I tend to get suspicious when I see a friend’s skull spontaneously explode.
w.r.t. the discussion about whether a robot could theoretically fire a rifle accurately at long-distance targets:
I’ve worked on AI systems that have needed to do similar (though non-lethal) tasks to sniping. Compensating for wind, air pressure, target motion etc is the kind of thing that many modern military systems must do. Accurate sniper fire is definitely achievable; I’m not saying it would ever be possible to be 100% accurate, nor that such a system could be more accurate than a human sniper.
But the notion that this problem would be intractable for modern automated systems is old-fashioned.
If robots haven’t been developed for sniping missions, I would assume it’s primarily because of all the extra decisions that need to be made in addition to simply taking the shot (friend or foe, which target to go for, whether now is the right time to fire etc).