Multi-Objective Particle Swarm Optimisation for Pareto Optimal Tuning of Super-Twisting Controller in Quadcopter

This work introduces the Multi-Objective Particle Swarm Optimisation (MOPSO) algorithm to optimise and speed up adjusting gains of a modified super-twisting second-order sliding-mode (MSTSOSM) attitude controller designed for a quadcopter. The tuning process is executed in MATLAB/Simulink, utilising a non-linear model that incorporates the physical characteristics of the quadcopter. The optimised parameters correspond to the MSTSOSM gains associated with each attitude state. The MSTSOSM constructs vectors using the gains of each Euler angle, employing particle coordinates computed by the MOPSO algorithm. These coordinates dynamically evolve during the algorithm's execution, exploring the parameter space for optimal positions. Consequently, the Pareto front emerges, showcasing Pareto's optimal solutions. The optimisation methodology is grounded in Pareto dominance theory, evaluating performance indicators such as rise time, root-mean-square error, and overshoot across all attitude axes in step responses. This systematic approach transcends traditional trial-and-error methods, comprehensively exploring the parameter space. Simulation and experimental results affirm the efficacy of this tuning process, demonstrating its capacity to enhance quadcopter control with precision and efficiency.