Multiobjective evolution of trajectories as multiple Bézier curves for UAVs
The SUAVs were initially designed for military applications, but they are becoming
more popular in civil applications due to technological advancements in cost reduction
and miniaturization. However, SUAVs’ autonomy and maneuverability are limited
because of their kinematics and low energy capacity constraints. Therefore, these
limitations must be taken into account when planning feasible trajectories for these
vehicles. The trajectories must attend the SUAVs’ kinematic constraints, visit multiple
waypoints and have minimized length. Currently, methods are capable of finding feasible
trajectories for SUAVs that attend the kinematic constraints, but cannot minimize length,
or visit multiple waypoints. This dissertation introduces TEvol, an algorithm capable of
finding feasible trajectories that attend to kinematic constraints of curvature, torsion and
climbing for multiple waypoints with minimized length. Furthermore, TEvol can also
consider kinematic constraints of velocity and load factor based on the specific SUAVs’
V-n diagram, which can protect the vehicle from structural failure and stall. The
trajectories are modeled as Bézier curves and optimized by a non-dominated sorting
genetic algorithm (NSGAII). The results indicate that the algorithm is capable of finding
trajectories with minimized length that visit multiple waypoints and attend to the
kinematic constraints of curvature, torsion and climbing, or velocity and load factor, and
both simultaneously. The algorithm found valid trajectories for 81.66% of the
experiments executed.