Field Deployment of a Plume Monitoring UAV Flock

Abstract

This letter describes and validates the robotic platform of an atmospheric plume monitoring system. The platform consists of a networked flock of unmanned aerial vehicles (UAVs) equipped with environmental sensors. The sensor flock forms an integral component of a dynamic data-driven application system (DDDAS) for plume monitoring. The goal of DDDAS is to dynamically incorporate data into a running simulation while simultaneously using the simulation to steer the measurement process. This letter takes a model-based approach to plume monitoring. From concentration measurements provided by the UAVs, a nonlinear online parameter estimator determines the model parameters. Based on the current knowledge of the model parameters, a hotspot identification routine directs the UAVs to information rich locations, or hotspots. The feasibility of deploying the testbed in uncontrolled outdoor environments is demonstrated with a flight test where three autonomous vehicles trace a simulated plume using simulated sensors.