Efficient Path Planning for UAVs to Recognize Chimneys With Excessive Exhaust Emissions

The severity of air pollution has prompted increased attention to air pollution monitoring. This paper uses unmanned aerial vehicles (UAVs) to discover chimneys with excessive exhaust emissions (called 3E-chimneys) in an industrial area. An efficient path-planning framework is proposed to find the flight path of a UAV for checking out chimneys and determine sampling points (SPs) to collect data. The framework builds the shortest path to visit chimneys. Four schemes are designed to select SPs around each chimney. The dual-location sampling (DLS) scheme chooses two SPs for each chimney according to its height. The upward spiral sampling (USS) scheme lets the UAV spiral upward at each chimney to gather data. In the inverted-U sampling (IUS) scheme, the UAV flies up to visit downwind SPs and then flies down to pass upwind SPs. The asynchronous isometric sampling (AIS) scheme picks an upwind SP and multiple downwind SPs. On doing the monitoring task, the UAV checks if some chimneys are skippable (i.e., not 3E-chimneys) using the industrial source complex (ISC3) model. In this way, the UAV can expedite the monitoring task and save energy. Simulation results reveal that the AIS scheme strikes a good balance between cost and performance for the monitoring task. The cost is defined by the length of the UAV's flight path. Four metrics are adopted to evaluate the performance: accuracy, recall, precision, and F1-score. Moreover, we make a prototype system to show the feasibility of our framework, which measures the concentration of CO2 gases emitted from small chimneys in a micro-field with a single wind direction to recognize 3E-chimneys.