This thesis presents the development and validation of a number of robust algorithms for both single and multi-quadrotor control for use in the nuclear industry, to improve the safety and efficiency of environmental monitoring processes associated with nuclear decommissioning. The objectives include the development and validation of a robust and reliable control algorithm for a single quadrotor, the design, development and validation of a robust multi-quadrotor system, and validation of the efficacy of the proposed control systems for applications in the nuclear industry.
A novel finite-time integral sliding mode control system was developed for robust trajectory tracking of a single quadrotor. This control method demonstrated superior performance over existing techniques in simulations and real-world experimentation in the presence of parameter uncertainties. A discrete-time sliding mode control system was also introduced to handle for the discrete nature of onboard sensors when controlling the quadrotors in environments where GPS is unavailable, showing enhanced performance when sampling rates were slow.
A discrete-time sliding mode formation control system was designed and implemented to enable the robust formation control of multiple quadrotors around a dynamic virtual leader in the presence of external disturbances. The efficacy of the algorithm was validated experimentally, through implementation on the Crazyflie 2.1 micro-quadrotor platform.
Finally, the applications of the proposed control algorithms for the nuclear industry was demonstrated by adapting the discrete-time sliding mode formation control with a gradient-climbing virtual leader, for locating the source of a radiative sensor field. Additionally, the data collection capability of the multi-quadrotor system was verified experimentally, using Gaussian Process Regression to estimate the temperature distribution within an environment. This highlighted the potential for the system to safely and efficiently monitor hazardous nuclear environments.
Overall, this thesis advances the field of quadrotor control by delivering robust algorithms tailored for the nuclear industry.