Minimum Distance and Minimum Time Optimal Path Planning With Bioinspired Machine Learning Algorithms for Faulty Unmanned Air Vehicles

Unmanned air vehicles operate in highly dynamic and unknown environments where they can encounter unexpected and unseen failures. In the presence of emergencies, autonomous unmanned air vehicles should be able to land at a minimum distance or minimum time. Impaired unmanned air vehicles define actuator failures and this impairment changes their unstable and uncertain dynamics; henceforth, path planning algorithms must be adaptive and model-free. In addition, path planning optimization problems must consider the unavoidable actuator saturations, kinematic and dynamic constraints for successful real-time applications. Therefore, this paper develops 3D path planning algorithms for quadrotors with parametric uncertainties and various constraints. In this respect, this paper constructs a multi-dimensional particle swarm optimization and a multi-dimensional genetic algorithm to plan paths for translational, rotational, and Euler angles and generates the corresponding control signals. The algorithms are assessed and compared both in the simulation and experimental environments. Results show that the multi-dimensional genetic algorithm produces shorter minimum distance and minimum time paths under the constraints. The real-time experiments prove that the quadrotor exactly follows the produced path utilizing the available maximum rotor speeds.