[DickingAround]

I don't think they can afford to see every bird as a threat during a battle. That means they'll also see every piece of shrapnel as a threat. And every piece of outgoing large munitions. I got the impression that speed and direction of travel are common filters. So something that's slow, tiny, and changing direction a lot is legitimately going to be hard to distinguish from a lot of non-threatening things.

[krapht]

I can't speculate on what's actually on US Navy warships right now, but I'm telling you commercially you can distinguish birds from drones under a kilometer away today. The knowledge has been around on how to do so with radar since 2006 https://ieeexplore.ieee.org/document/1603402, but even without that, there are numerous combined radar-camera surveillance systems out there for accurate target classification. Not like you don't have time to cue a camera to look at an interesting radar target when it's travelling slowly like a UAV might.

[runjake]

The military’s been able to distinguish this since the early 90s — at least. And yes modern drones didn’t exist in the 90s but other similarly-sized machines did, eg balloon-mounted telemetry devices.

Citation: personal experience with those sensor systems.

[jandrewrogers]

The parent is correct. They have considerable capacity to accurately classify everything in their regional vicinity in milliseconds. This is old capability.

Tracking millions of entities in your general space in real-time is definitely possible today. I have little reason to believe the US Navy cannot do it too.

[nonameiguess]

Unless wildlife is wearing armor, the radar and even polarimetric light signature is vastly different between a bird and shrapnel even if it's not moving at all. Not all detection methods are heuristic.