Sound carries information that visual sensors miss, and it does so passively — no active signal is required, and detection range can exceed line of sight. Identifying and roughly localising an acoustic event of interest (an explosion, a vehicle, a structural failure) from a set of low-cost distributed microphones is well-established in principle; making it work in real time, on constrained hardware, in a noisy and reverberant real-world environment is where it gets harder.

Under fire or in any acute field incident, voice reporting of where a threat came from is slow, stressful, and error-prone. A set of small microphones distributed across a team — worn or vehicle-mounted — can, in principle, work out the direction and rough distance of a loud, sudden acoustic event automatically, using differences in arrival time between microphones (time-difference-of-arrival, or TDOA, localisation).

How accurately can a small, distributed microphone network estimate the direction and approximate distance of a defined class of loud, sudden acoustic events — and how should that estimate be presented to a team leader as a clear, actionable information on a shared map?

In this project we aim to implement a detection and TDOA localisation solution and designs the map-overlay interface for the leader view.

Join the project if you're interested in signal processing, embedded systems, acoustics, or turning a noisy real-time sensor stream into a fast, actionable signal for someone under pressure.