ATMOS is a multi‑domain early warning platform that materializes invisible threats in air by monitoring the coupled behavior of gases, aerosols, humidity and kinetic signatures in real time. In Phase 1 (ATMOS‑BIO), the system is piloted in high‑humidity urban environments using safe tracer odours, CO₂ and multi‑gas sensing to generate real‑time “aerosol loading maps” that reveal stagnant air pockets where exhaled breath and virus‑laden particles accumulate, validating the core tracer‑based risk model. In Phase 2 (ATMOS‑MINE), the same sensing engine is extended to underground mines, where methane, humidity and dust are monitored alongside accelerometers and geophones; this enables detection of both grizu pre‑threshold conditions and abnormal vibration patterns consistent with structural failure or side‑channel sabotage. By fusing gas/aerosol anomalies with kinetic anomalies through a dual‑condition logic, ATMOS can trigger automated protective responses—such as industrial lockdown and emergency ventilation—before a spark or mechanical event escalates into a catastrophic explosion.

PHASE 1: ATMOS‑BIO – Urban Pilot and Tracer Validation
Objective:
To validate the correlation between “Tracer Odour” plumes and aerosol accumulation in high‑humidity urban environments.
1.1 Experimental Design (Tracer–Odour Coupling)
Environmental Seeding:
Controlled release of a non‑toxic, easily detectable organic tracer (e.g., Ethyl Acetate or a food‑grade synthetic scent) into a designated “Infection Risk Zone” such as a subway entrance, an underpass, or a congested street corner.
High‑Humidity Campaigns:
Pilot deployments will be conducted during periods with Relative Humidity (RH) > 80% to capture the “aerosol‑trapping” regime, in which droplets do not fully evaporate but instead remain suspended and bind to tracer molecules.
1.2 Sensor Array Deployment (The Detection Grid)
Multi‑Gas Sensor Network:
Deployment of a multi‑point sensor grid using MOS (Metal‑Oxide Semiconductor) and PID (Photo‑Ionization Detector) units to map the spatial and temporal dispersion of the tracer plume in three dimensions.
Baseline and Fingerprinting:
Establishment of a robust olfactory baseline for the test area by distinguishing background urban odours (exhaust, waste, food, industrial emissions) from the injected “Tracer Fingerprint” using edge‑deployed AI/ML for real‑time pattern recognition and noise rejection.
1.3 Risk Mapping and Visualization
ATMOS Real‑Time Heat Map:
Development of the ATMOS Real‑Time Heat Map Engine, which converts gas concentration and tracer intensity data into an Aerosol Loading Map highlighting stagnant or poorly ventilated zones.
Correlation and Validation:
Continuous co‑measurement of CO₂ concentrations and environmental parameters (temperature, RH, air‑flow) to validate the core hypothesis: zones where the tracer odour accumulates and persists are also zones where exhaled breath (CO₂) and virus‑laden aerosols accumulate and persist. This correlation will be used to derive risk indices for infection‑prone micro‑environments.
PHASE 2: ATMOS‑MINE – Kinetic Integration and Sabotage Prevention
Objective:
To extend the ATMOS platform to industrial safety by integrating seismic/vibrational sensing, enabling early detection of structural risks and potential side‑channel sabotage in underground mines.
2.1 The “Methane Sponge” Calibration
CH₄–Humidity Characterization:
Calibration of the sensor suite to detect the characteristic signature of Methane (CH₄) absorption and retention in humid mine air, including interaction effects with water droplets and dust particles.
Grizu Pre‑Threshold Detection:
Training of the ATMOS AI models to identify the Pre‑Saturation Point—the condition where the combination of methane concentration and humidity approaches a critical ratio associated with grizu (firedamp) explosion risk, even before classical threshold alarms are reached.
2.2 Kinetic Side‑Channel Integration
Vibration and Micro‑Seismic Instrumentation:
Installation of high‑sensitivity accelerometers and geophones on critical mine infrastructure: elevator shafts, rail tracks, conveyor systems, main support pillars, and key junctions.
Operational vs. Sabotage Signatures:
Development of a vibration signature library that separates normal operational patterns (e.g., routine wagon movement, equipment start/stop cycles) from anomalous kinetic profiles indicative of targeted mechanical strain, deliberate overloading, or unauthorized manipulation consistent with side‑channel sabotage scenarios.
2.3 Cross‑Verification Logic (The “Smart‑Kill” Layer)
Dual‑Condition Alert Engine:
Implementation of a dual‑condition logic within ATMOS‑MINE, based on:
Condition A – Kinetic Anomaly: Localized detection of abnormal vibrational or micro‑seismic activity (potential sabotage, incipient collapse, or structural failure).
Condition B – Gas/Aerosol Anomaly: Simultaneous detection of gas stagnation, abnormal methane build‑up, or “plume freezing” (trapped, high‑humidity air) within the same spatial sector.
Automated Protective Response:
When both conditions are satisfied within a defined time‑window, the system escalates to Critical Alert, triggering:
Immediate industrial lockdown of the affected sector (halt of ignition‑capable machinery, controlled power‑down).
Forced emergency ventilation and, where available, inertisation protocols.
Automatic notification to control room operators and safety teams, with spatially resolved diagnostic data to enable rapid decision‑making before a catastrophic explosion can be triggered.
Project Milestones – Phases 1 & 2
M1 (Month 6):
Demonstrated detection and mapping of a controlled tracer plume over at least a 50‑meter urban radius under RH > 80% conditions, including validated correlation between tracer persistence, elevated CO₂, and stagnant‑air micro‑zones.
M2 (Month 12):
Successful detection of a simulated side‑channel kinetic attack in a controlled mining testbed, through the combined analysis of vibration signatures and gas/aerosol stagnation patterns, leading to a correctly triggered dual‑condition alarm without false positives during normal operation.