The platform

Project ATLAS

A modular, drone-based sensing platform for high-resolution, three-dimensional characterisation of the lower atmosphere. Designed to expand progressively alongside research ambitions.

Overview

Built for the atmosphere.

ATLAS combines physics-informed modelling with machine learning to reconstruct three-dimensional environmental fields from sparse measurements. This allows investigations of temperature gradients, airflow dynamics, pollutant dispersion, and transient atmospheric phenomena that would be impractical with traditional fixed or point-based instruments.

The platform is designed around a modular sensor architecture — instrumentation is selected and combined according to research objectives, payload constraints, and environmental conditions. Capability is added incrementally, only after the preceding configuration has been fully validated.

This staged approach ensures technical robustness, manageable payload constraints, and flexibility across multiple research directions.

Key capabilities

  • →3D thermal mapping & vertical profiling
  • →Atmospheric stability & stratification analysis
  • →Flow field estimation from IMU & anemometry
  • →ML-based volumetric field reconstruction
  • →Pollutant dispersion & source localisation
  • →Transient & extreme event capture

Sensor architecture

Modular instrumentation.

ATLAS integrates a range of environmental, atmospheric, and imaging sensors to enable multi-domain characterisation of the lower atmosphere.

Sensor Type / Instrument Measures Research applications
Temperature High-res digital thermistor Air temperature Vertical thermal profiles, urban heat island mapping, stability analysis
Humidity Relative humidity sensor RH, dew point Fog formation studies, microclimate mapping, moisture flux
Barometer Pressure sensor Air pressure, altitude Precise altitude estimation, pressure gradient detection, pre-event weather signatures
IMU Inertial measurement unit Acceleration, angular velocity, orientation Turbulence intensity estimation, gust detection, wind disturbance inference
GPS GNSS receiver Position, velocity over ground Geospatial mapping, field reconstruction, wind estimation via ground/air speed differential
PM2.5 / PM10 Optical particle counter Fine particulate concentration Pollution plume mapping, dispersion visualisation, exposure mapping
Gas sensors Electrochemical / NDIR CO₂, NO₂, O₃ Urban emission analysis, transport modelling, source localisation
Anemometer Ultrasonic anemometer 3D wind velocity Turbulence spectra, wind shear detection, flow visualisation, vorticity estimation
Pressure probe Multi-hole probe (pitot system) Airspeed, flow direction Direct flow angle measurement, complementary to anemometry
Thermal camera IR imaging sensor Surface temperature Heat leakage mapping, urban heat visualisation, thermal plume detection above structures

Sensor stacks

Progressive capability.

ATLAS employs a staged architecture. Each stack builds on the last, and is only deployed following successful validation of all preceding configurations.

Stack 1 Core Atmospheric Profiling Current
Temperature Humidity Barometer IMU GPS
  • Vertical temperature gradient analysis
  • Atmospheric stability assessment
  • Urban heat island mapping
  • Basic turbulence inference from IMU disturbances
  • 3D thermal field reconstruction via interpolation

Lightweight · low power · high reliability · ML reconstruction baseline

Stack 2 Environmental Dispersion & Air Quality
Stack 1 PM2.5 / PM10 CO₂ / NO₂ / O₃ (opt.)
  • Pollution plume mapping
  • Advection–diffusion modelling
  • Source localisation attempts
  • Time-dependent dispersion reconstruction
  • ML-based spatial interpolation of sparse pollutant samples

Trade-off: increased calibration complexity, slower sensor response times

Stack 3 Flow Field Characterisation
Stack 1 Ultrasonic Anemometer OR Multi-Hole Probe
  • 3D wind vector mapping
  • Turbulence intensity estimation
  • Wind shear detection
  • Vorticity approximation
  • Validation of CFD or simplified flow models

Trade-off: higher cost, increased payload, sensitive mounting requirements

Stack 4 Thermal Imaging & Surface Interaction
Stack 1 Thermal Camera
  • Surface heat flux estimation
  • Urban heat trapping visualisation
  • Thermal plume detection above structures
  • Building-envelope heat leakage mapping

Enables surface–atmosphere interaction studies at urban scale

Stack 5 Adaptive Intelligence Platform Future
Any Stack 1–4 Onboard Edge Computing
  • Real-time gradient detection
  • Adaptive waypoint adjustment
  • Uncertainty-driven sampling
  • ML-informed path optimisation

Operational constraints

Physical realities.

ATLAS operates within the physical limits of a small unmanned aerial platform. Sensor selection and stack configuration are governed by the following constraints.

Payload mass

Each additional sensor increases total weight, affecting thrust requirements, motor loading, and flight endurance. Payload must remain within safe takeoff limits.

Power consumption

Active sensors and onboard processing draw from the flight battery. Increased electrical load reduces flight time and may introduce thermal management challenges.

Flight endurance

Battery capacity limits total mission duration. Heavier configurations reduce achievable altitude range, profiling depth, and spatial coverage.

Aerodynamic disturbance

External probes alter airflow around the drone body. Sensor placement must minimise rotor wash interference and structural vibration.

Sensor response time

Gas and particulate sensors have slower response times. Rapid flight through gradients can introduce spatial lag in measurements.

Data bandwidth

High-frequency sensors and imaging systems generate significant data volumes. Logging speed, storage, and synchronisation accuracy constrain acquisition strategies.