4 ms·

The Israeli’s Iron Curtain is an example of a possible defensive model. Computer controlled machine guns running ML models could probably take down a lot of dro

The Israeli’s Iron Curtain is an example of a possible defensive model. Computer controlled machine guns running ML models could probably take down a lot of drones.

jmpman 7y ago
Iron Curtain only works for objects that are detected. If the drones are flying at 50 feet AGL, Iron Curtain isn’t going to work as they’re not going to detected by radar, and exploding an Iron Curtain interceptor 50 feet above someone’s house isn’t going to go well. Flying at 10,000 ft, how far away can an RC motor be detected and not look like a bird? Iron Curtain currently has 10-15 batteries at four launchers per battery with 20 interceptors per launcher. So, 800-1200 interceptors. With only 1200 interceptors, they can be overwhelmed by distributed terrorists launching simultaneously over a 1000 mile diameter. Each interceptor is $40k. I expect a drone would be under $500, so you’ve got an 80:1 cost differential. $400M to stop 10,000 drones. As for machine guns, let’s say the effective range is 2000 yards. You have roughly 60 seconds to engage all targets. Assume 100 rounds fired to successfully terminate a single drone, with 10,000 drones inbound, you need to fire 1M bullets in 60 seconds. Assuming a machine gun firing at 1000rds/min, that’s 1000 machine guns with perfectly coordinated fire. At $10k/machine gun (a very conservative estimate) how many buildings can you protect for $10M each? Maybe you can protect your capital building with such a lead shield. Your oil refineries.. should probably do that. And now you’ve just sprayed 1M bullets into the air above an oil refinery. Sounds like a bad idea. And then the terrorist just need to attack with 20,000 drone swarm, and all your facilities are exposed until you double your defenses at all locations. Financially, I don’t see a solution.
- sjtindell 7y ago
  Good points. Your answer would pass the systems design interview and mine would not.
  - jmpman 7y ago
    A laser however... probably works for this scenario. But even then, there’s a required dwell time, and more sophisticated terrorists would adapt to using ablative skins, or maybe even materials transparent to the laser wavelength... anything to increase the required dwell time and overwhelm through sheer numbers.
    - sjtindell 7y ago
      Perhaps the best way would be jamming or hijacking the signal. You could essentially "remove" huge groups of drones at a time.
      - jmpman 7y ago
        In order to do that, the electronics inside the drone would need to be susceptible to the electronic noise. Sure, RF shielding from a military grade perspective is difficult, but for this purpose, copper paint surrounding the electronics enclosure may be sufficient. With GPS denied, inertial and visual navigation would be required. Inertial navigation quickly succumbs to accumulated error, but if GPS signals are stopped (jammed) just a few minutes before impact, INS will probably work, given a calm atmosphere, there won’t be much noise inducing turbulence. Visual navigation is much harder, but the advances in machine learning are making this problem trivial. RF interference is likely the best defense, as quality RF shielding is difficult, especially at extreme frequencies.
- Fjolsvith 7y ago
  AI will be developed to spot them, I'm sure.

The Israeli’s Iron Curtain is an example of a possible defensive model. Computer controlled machine guns running ML models could probably take down a lot of drones.

jmpman 7y ago

Iron Curtain only works for objects that are detected. If the drones are flying at 50 feet AGL, Iron Curtain isn’t going to work as they’re not going to detected by radar, and exploding an Iron Curtain interceptor 50 feet above someone’s house isn’t going to go well. Flying at 10,000 ft, how far away can an RC motor be detected and not look like a bird? Iron Curtain currently has 10-15 batteries at four launchers per battery with 20 interceptors per launcher. So, 800-1200 interceptors. With only 1200 interceptors, they can be overwhelmed by distributed terrorists launching simultaneously over a 1000 mile diameter. Each interceptor is $40k. I expect a drone would be under $500, so you’ve got an 80:1 cost differential. $400M to stop 10,000 drones. As for machine guns, let’s say the effective range is 2000 yards. You have roughly 60 seconds to engage all targets. Assume 100 rounds fired to successfully terminate a single drone, with 10,000 drones inbound, you need to fire 1M bullets in 60 seconds. Assuming a machine gun firing at 1000rds/min, that’s 1000 machine guns with perfectly coordinated fire. At $10k/machine gun (a very conservative estimate) how many buildings can you protect for $10M each? Maybe you can protect your capital building with such a lead shield. Your oil refineries.. should probably do that. And now you’ve just sprayed 1M bullets into the air above an oil refinery. Sounds like a bad idea. And then the terrorist just need to attack with 20,000 drone swarm, and all your facilities are exposed until you double your defenses at all locations. Financially, I don’t see a solution.

sjtindell 7y ago
Good points. Your answer would pass the systems design interview and mine would not.
- jmpman 7y ago
  A laser however... probably works for this scenario. But even then, there’s a required dwell time, and more sophisticated terrorists would adapt to using ablative skins, or maybe even materials transparent to the laser wavelength... anything to increase the required dwell time and overwhelm through sheer numbers.
  - sjtindell 7y ago
    Perhaps the best way would be jamming or hijacking the signal. You could essentially "remove" huge groups of drones at a time.
    - jmpman 7y ago
      In order to do that, the electronics inside the drone would need to be susceptible to the electronic noise. Sure, RF shielding from a military grade perspective is difficult, but for this purpose, copper paint surrounding the electronics enclosure may be sufficient. With GPS denied, inertial and visual navigation would be required. Inertial navigation quickly succumbs to accumulated error, but if GPS signals are stopped (jammed) just a few minutes before impact, INS will probably work, given a calm atmosphere, there won’t be much noise inducing turbulence. Visual navigation is much harder, but the advances in machine learning are making this problem trivial. RF interference is likely the best defense, as quality RF shielding is difficult, especially at extreme frequencies.
Fjolsvith 7y ago
AI will be developed to spot them, I'm sure.