I thought autonomous killbots were still not used (officially,) outside like Israel. There is some agreement not to go full autonomous, not sure the signatories. Anyone have any confirmation on this?
They’re saying two different things in the post. One is that a drone is flown to a location and then automatically attacks whatever it encounters, that’s fully automatic and would be aweful imo because then the drone is making decisions on what to attack/kill.
The second thing is later in the post when they talk about countermeasures. They state that the drones can be countered before they’re locked on to a target. That implies that an operator picks a target and the automation part only serves to keep the drone locked on the target, but the drone makes no decision on what to target. That’s conceptually similar to a pilot selection an enemy aircraft they locked onto before firing a missile.
From the post it’s not quit clear which of the two are the case or if both are happening. I suspect it’s the second case because Ukraine can’t really afford the bad publicity of the first case imo.
It’s true, locking onto a target is no different than a guided missile. The Russian blogger is trying to make it sound like it’s autonomous killbots, when it’s guided munitions that are set on a target, by people, to avoid jamming.
Local onboard ‘AI’ is used for image recognition, classification and target locking usually, and it flies using essentially agrodrone software missions (ardupilot). It’s not completely autonomous, just mostly hands off.
It does sound like someone clicks the right blob onscreen, so in the absence of further instruction, the drone will go give it a hug. That’s missile lock-on via webcam, not a flying landmine.
I thought autonomous killbots were still not used (officially,) outside like Israel. There is some agreement not to go full autonomous, not sure the signatories. Anyone have any confirmation on this?
They’re saying two different things in the post. One is that a drone is flown to a location and then automatically attacks whatever it encounters, that’s fully automatic and would be aweful imo because then the drone is making decisions on what to attack/kill.
The second thing is later in the post when they talk about countermeasures. They state that the drones can be countered before they’re locked on to a target. That implies that an operator picks a target and the automation part only serves to keep the drone locked on the target, but the drone makes no decision on what to target. That’s conceptually similar to a pilot selection an enemy aircraft they locked onto before firing a missile.
From the post it’s not quit clear which of the two are the case or if both are happening. I suspect it’s the second case because Ukraine can’t really afford the bad publicity of the first case imo.
It’s true, locking onto a target is no different than a guided missile. The Russian blogger is trying to make it sound like it’s autonomous killbots, when it’s guided munitions that are set on a target, by people, to avoid jamming.
It’s probably not true that they’re using AI for this.
Local onboard ‘AI’ is used for image recognition, classification and target locking usually, and it flies using essentially agrodrone software missions (ardupilot). It’s not completely autonomous, just mostly hands off.
It does sound like someone clicks the right blob onscreen, so in the absence of further instruction, the drone will go give it a hug. That’s missile lock-on via webcam, not a flying landmine.