Real-time prediction of gas contaminant concentration from a ground intruder using a UAV

This work presents a combined computational fluid dynamics-controls approach that guides an unmanned aerial vehicle (UAV) in a region of a plume and provides real-time estimates of gas concentration from a ground mobile source with unknown location and release rate. The contaminant concentration field is estimated in real-time on the UAV via the use of a state estimator that is based on a Luenberger observer design. UAV-affixed concentration sensors provide measurements which are used by the controls scheme to guide the UAV into regions of useful information in order to expedite the estimation of the contaminant concentration field. The estimator is solved in real-time using a finite-volume with total-variation-diminishing scheme on adapted grids. The UAV guidance scheme is coupled to the estimator performance through Lyapunov redesign methods, according to which the UAV is guided towards areas of larger estimation error. Numerical examples illustrate the performance of the scheme for different atmospheric conditions and mobile contaminant sources.