Flight Controller Optimization of Unmanned Aerial Vehicles using a Particle Swarm Algorithm

In this paper, a simultaneous calibration algorithm of the parameters of the attitude and altitude control for an unmanned aerial vehicle (UAV) is proposed. The algorithm is based on the multi-objective particle swarm optimization (MOPSO) technique. This algorithm is implemented by using the free PX4 software for the Pixhawk2 controller. The behavior of the UAV is simulated given its physical characteristics by means of a non-linear model and a search of the controller parameters. This latter is based on a proportional (P) position controller in cascade with a proportional-integral-derivative (PID) speed controller of its height and each one of its Euler angles. To perform this search, the PID gains K
_p1
, K
_p2
, K
_i
and K
_d
of each of the degrees of freedom are used to define vectors considered particle positions by the MOPSO algorithm, which moves them through a search space to find sets of optimum values according to Pareto, or the Pareto Front. The search is carried out based exclusively on Pareto dominance concepts, comparing parameters of step responses (overshoot, rise time, root-mean-square error) of each of the degrees of freedom. In order to show the efficiency of the proposal, simulation results are provided by using the calibration methodology obtaining good results.