Drone-Sonde Development, Modelling, and Control for Atmospheric Measurements

Abstract

This PhD focuses on assessing the capability of multirotor vehicles as a replacement for conventional balloon-launched radiosondes in meteorology. Compared with radiosondes, the most challenging aspect of sensing with multirotors was wind speed and direction. Several approaches to measuring the wind speed and direction based on the vehicle's response were compared in hover conditions, including a novel method that requires only flight data for model development.

The accuracy of these models was assessed against each other and the World Meteorological Organisation's accuracy requirements for wind measurement producing mean absolute errors with a range of 0.29m/s to 2.39m/s, depending on the method additionally, the direction accuracy was compared and found to be outside of the 5.00$^\circ$ requirement, with a mean absolute error of 10.19$^\circ$.

The wind estimation methods were further developed to consider flight outside of the hover condition by correcting for the velocities and accelerations of the vehicle. This correction was validated against data gathered using a sonic anemometer mounted above the vehicle. The results show a mean error ranging from 0.36m/s to 0.52m/s across climb rates from 2m/s to 5m/s. This estimation method was used alongside commercially available radiosondes for atmospheric vertical profiling in Guatemala, reaching altitudes over 4000m above mean sea level.

Once an accurate approach to wind estimation was developed, methods to improve the endurance and safety of operations were considered based on the measured wind conditions. An approach to optimal path generation was investigated while considering the lateral airspeed for stable descent, showing that it is possible to reduce energy consumption in descent under known wind conditions, achieving a maximum energy saving of 67.51\%.

Taking the next steps to consider a varying wind environment, we consider an approach to trajectory and airspeed management based on predictive forecasting of changes in wind speed and direction to allow dynamic rerouting based on the changing environment for energy saving. Producing small energy savings of 2.93\% under low wind speed conditions.

Date of Award	19 Mar 2024
Original language	English
Awarding Institution	University of Bristol
Supervisor	Tom S Richardson (Supervisor) & Arthur G Richards (Supervisor)