The cost curve problem

Counter-UAS is usually discussed as a sensing and effects problem. It is more usefully understood as an arithmetic problem, because the arithmetic is what makes it urgent.

The exchange rate

If a threat can be produced for a small sum and the established method of stopping it costs orders of magnitude more, the defender loses the exchange even while winning every engagement. Sustained against numbers, that is not a tactical inconvenience — it is a resourcing failure.

What makes small uncrewed aircraft genuinely awkward is that they sit precisely in the gap between existing categories. They are too small and slow for systems designed against fast aircraft, too numerous for engagement methods that assume a handful of targets, and too cheap for the exchange rate to work in the defender's favour.

You can win every engagement and still lose, if each win costs more than the thing you destroyed.

Detection is harder than it sounds

Before anything else, you have to see it — and small aircraft are legitimately difficult to detect. They are physically tiny, fly low and slowly, are frequently made of materials that return little to conventional sensing, and appear against cluttered backgrounds of terrain, buildings and vegetation.

Then there is the discrimination problem. At useful range, distinguishing a small aircraft from a bird is not trivial, and getting it wrong in either direction is costly: miss the aircraft and the system has failed; flag every bird and the system is switched off within a week.

• Small physical size at operationally useful ranges.
• Low and slow flight profiles that many sensing approaches were not designed for.
• Cluttered backgrounds rather than open sky.
• Multiple simultaneous tracks rather than one.
• The need to identify, not merely detect.

Where autonomy actually helps

Our interest is deliberately confined to the sensing and understanding half: reliably noticing something is there, holding a track on it as it manoeuvres, handling several at once, and working out what it is — fast enough that the warning still has value.

This is a perception problem of the hardest kind, which is why it connects directly to the rest of our work. Small, distant objects in clutter, in poor conditions, with a very low tolerance for false alarms, running on constrained compute. Improvements here benefit every other application we build for.

Where we stop

What happens after identification — whether and how to respond — is a decision for the human commanders responsible for it. We build the picture; people decide what to do about it. That boundary follows directly from our position that a person is always in command, and we are explicit about it because in this application the distinction matters more than most.

The honest position

Counter-UAS is not solved, by us or by anyone. It is a genuinely hard sensing problem attached to a genuinely hard economic one, and the threat side is improving quickly and cheaply. Our systems remain in research and development.

What we can say is that the shape of a workable answer is fairly clear: detection that holds up in real conditions, discrimination good enough to be trusted, and a cost structure that does not lose the exchange. Getting all three at once is the work.