ARPAS UK Bird&Bird UAS Roundtable

At the ARPAS-UK and Bird & Bird UAS Roundtable, Droneod's contribution was to raise the Notting Hill Carnival context: what happens when drones are deployed over a two-million-person public event, and what the sensing stack behind that deployment is actually built on.

The reporting that followed bore the point out. The British Transport Police's first dedicated Carnival drone team is a capable deployment — wide-angle and zoom optics, thermal imaging for low-light and concealment, a laser rangefinder generating GPS coordinates at over a kilometre, feeds routed live to officers and a CCTV hub, and live facial recognition running alongside. But the entire detect-identify-locate chain rests on what a camera can see. And the facial recognition layer drew criticism from civil rights groups over uneven accuracy across demographic groups — a reminder that imagery-based AI carries bias, not merely blind spots.

That is the pattern Droneod flagged. When the sensing stack is built around imagery, its assumptions are inherited wholesale by everything downstream — including counter-UAS, where the same AI-driven radar and electro-optical dependency is becoming the default for threat detection and classification