AetherWeave

Inspiration

When a metal bird falls from the sky, it vanishes into the unforgiving embrace of the earth. The immediate human response—scrambling helicopters, squinting through binoculars, and manually plotting grids—is tragically archaic.We were inspired by the mathematical paralyzation that occurs during these critical hours. If an aircraft's last known velocity is $v$, and time $t$ has elapsed, the search area expands geometrically: $$A = \pi (v \cdot t)^2$$ Factoring in wind drift vectors ($\vec{w}$) and terrain elevation maps, the probability distribution of the crash site requires continuous, real-time Bayesian updating. Relying on slow, centralized human coordination to traverse this expanding abyss is a fatal bottleneck. We envisioned a symphony of synthetic minds: a swarm that does not sleep, does not panic, and communicates flawlessly to blanket the search zone.

How we intent to built it

We engineered a Multi-Agent System (MAS) decoupled through an event-driven architecture.First, we built an ingestion engine that pulls live telemetry from the OpenSky Network and wind data from NOAA. This data feeds into our orchestrator node, which calculates the highest-probability crash polygon.Instead of deploying a monolithic script, we containerized the search logic. The orchestrator spawns independent, lightweight worker agents, assigning each a specific geographic bounding box. We bypassed slow HTTP requests and wired the swarm's nervous system using a high-throughput message broker.The probability of finding the target in a specific grid cell $i$ is constantly updated across the swarm using a simplified Bayesian model: $$P(\text{Found} \mid \text{Grid}_i) = P(\text{Crash in } i) \times (1 - P(\text{False Negative} \mid \text{Pass}))$$ When a worker drone agent detects a simulated thermal or visual anomaly, it instantly publishes the event to the broker. The surrounding agents, subscribed to the topic, dynamically abandon their empty grids and converge on the anomaly's coordinates in real-time. The entire chaotic dance is visualized on a clean, responsive frontend dashboard.