Most drone demos happen on a clear, sunny day. The hard part is everything else: heavy rain, glare, a low sun, and a camera that never sits still.
I'm testing a drone sensing system in simulation, on full city streets with traffic. We build the scenes, add the bad weather, and then try to break the system on purpose, scene after scene, until we find where it fails. The goal is simple to say and hard to do: drones that know when they can't be trusted.
This is the first in a series of short research teasers. I'll share more as the work moves along.