[iramiller]

How long until vision based ‘AI’ systems make these kinds of counter measures obsolete? It seems like it should be straight forward for a multi-modal sensor package to combine with some basic physics logic (the airframe can only ‘move’ so much) which would defeat these types of counter measures. The vision based processing component should be able to filter out chaff and flare based disruption.

I expect that the Ukraine war is a very useful proving/testing ground for a great many defense industry companies.

[tivert]

> How long until vision based ‘AI’ systems make these kinds of counter measures obsolete? It seems like it should be straight forward for a multi-modal sensor package to combine with some basic physics logic (the airframe can only ‘move’ so much) which would defeat these types of counter measures. The vision based processing component should be able to filter out chaff and flare based disruption.

And then the countermeasure can maybe just flash a weird pattern of light at the missile to confuse the ‘AI’ systems and it starts thinking the jet is a cloud (e.g. https://www.theregister.com/2017/11/06/mit_fooling_ai/, https://www.youtube.com/watch?v=qPxlhGSG0tc).

[wrp]

Some current systems already have trouble at night. I would expect vision based systems to have even more trouble. Are there any CV based guidance systems?

[z303]

The first to come to mind is the AGM-65 Maverick in both visible light and IR (https://en.wikipedia.org/wiki/AGM-65_Maverick). Later version of the AIM-9 Sidewinder also have IR imaging seekers (https://en.wikipedia.org/wiki/AIM-9_Sidewinder) as do the AIM-132 ASRAAM (https://en.wikipedia.org/wiki/ASRAAM). I'm sure there are a bunch of others.

[wrp]

Need to clarify here. GP appeared to be asking about computer vision based systems, which would involve object recognition. The traditional vision systems such as you referenced are based on contrast detection.