Coping with Rotor Wash Effect in Outdoor PM Measurements with an Autonomous UAV

While using Unmanned Aerial Vehicles (UAV) as air monitoring stations presents significant advantages over fixed stations, they present an important limitation which is the rotor wash effect, which compromises air sensors’ accuracy. This work proposes a UAV equipped with a sensor separation mechanism and evaluates the accuracy of measurements taken with a PAS-OUT-01 PM sensor placed in a UAV at different separation distances by comparing them with a fixed reference sensor. Linear models with Newey–West robust standard errors show that differences from the reference sensor are significant (p < 0.05) with high values on error metrics when the sensor is not separated from the UAV, indicating high disturbance showing MAAPE values of 1.55 rad for PM2.5 and 1.55 rad for PM10, Bias values of 86.85 µgm3 for PM2.5 and 91.00 \(µg/m^3 \) f or PM10, and RMSE of 131 \(µg/m^3\) for PM2.5 and 135 \(µg/m^3\) for PM10. No significant differences (p ≥ 0.05) and low error were found at a 7 m separation, indicating a high agreement showing MAAPE values of 0.054 rad for PM2.5 and 0.080 rad for PM10, Bias values of 0 for PM2.5 and −0.02 \(µg/m^3\) for PM10, and RMSE of 0.36 \(µg/m^3\) for PM2.5 and 0.91 \(µg/m^3\) for PM10. Intermediate comparisons do not show a linear decrease in the significance of the differences and error metrics. Additionally, flight tests showed that the UAV achieves a mean flight time of 14 min and a mean autonomous landing deviation of 1.55 m while carrying the proposed mechanism fully retracted.