I conducted my Master Thesis @ Applied Cognitive Psychology Lab at Ulm Universität using Optitrack motion tracker, I used to sit and observe a CNC-driven loudspeaker sweep past a subject's head 100's of time collecting rich quality motion tracking data.

My thesis tracked how people's heads move while following a moving sound source. The part I built that didn't exist before: a way to process head kinematic data efficiently enough to actually use it, refining a few studies that couldn't.

The result wasn't what I expected, but that's the beautiful things about science- You can always learn something with data. It supports the two-point snapshot model for auditory motion perception. That is we don't track moving sound continuously the way we track moving objects visually, we sample it in snapshots. Visual and auditory motion perception run on different algorithms, and any system trying to replicate either has to respect that.

That's exactly why it matters for drones. Vision fails in smoke, darkness, obstruction. Sound doesn't. Its application are wide and beyond, especially for drones. One use case for such is in law enforcement. Localizing gunfire properly means building the auditory side on its own terms, not applying vision models onto microphones.

Thesis: https://oparu.uni-ulm.de/items/6b2e607f-f048-4610-b13a-3c5a26c88591