Cooperative Autonomous Observation of Coherent Atmospheric Structures using Small Unmanned Aircraft Systems

Mapping the structure of localized atmospheric phenomena, from sea breeze and shallow cumuli to thunderstorms and hurricanes, is of scientific interest. Low-cost small unmanned aircraft systems (sUAS) open the possibility for autonomous "instruments" to map important small-scale phenomena (kilometers, hours) and serve as a testbed for for much larger scales. Localized phenomena viewed as coherent structures interacting with their large-scale environment are difficult to map. As simple simulations show, naive Eulerian or Lagrangian strategies can fail in mapping localized phenomena. Model-based techniques are needed. Meteorological targeting, where supplementary UAS measurements additionally constrain numerical models is promising, but may require many primary measurements to be successful.

We propose a new, data-driven, field-operable, cooperative autonomous observing system (CAOS) framework. A remote observer (on a UAS) tracks tracers to identify an apparent motion model over short timescales. Motion-based predictions seed MCMC flight plans for other UAS to gather in-situ data, which is fused with the remote measurements to produce maps. The tracking and mapping cycles repeat, and maps can be assimilated into numerical models for longer term forecasting.

CAOS has been applied to study small scale emissions. At Popocatepetl, in collaboration with CENAPRED and IPN, it is being applied map the plume using remote IR/UV UAS and in-situ SO2 sensing, with additional plans for water vapor, the electric field and ash. The combination of sUAS with autonomy appears to be highly promising methodology for environmental mapping.

For more information, please visit http://caos.mit.edu