Exploring atmospheric phenomena with a fleet of UAVs
The fine understanding or monitoring of atmospheric phenomena such as clouds or plumes requires the acquisition of spatially dense information over their lifespan. Fleets of fixed-wing UAVs carrying dedicated sensors are well suited for this purpose, but the size and duration of the phenomena to observe are such that no systematic sampling strategies can be implemented, as they would call for the coordination of dozens of UAVs. Instead, adaptive sampling strategies that reason in real time on the acquired data can efficiently gather the required information.
This talk will present on-going work on developing such adaptive schemes. It will in particular depict the way the acquired data are gathered into stochastic maps, which are exploited to drive the fleet. Preliminary results obtained during a first flight campaign targeted to the analysis of cumulus clouds will be analysed, and other issues related to such applications for UAVs will be briefly presented, such as optimal UAV conception and lightweight atmosphere sensor developments.