Collision-free path planning based on a genetic algorithm for quadrotor UAVs

Path planning is one of the most important topics for applications of UAVs. Genetic algorithms are minimization tools that are widely used to process large amounts of data. In this research, a genetic algorithm capable of generating navigation waypoints, achieving short distances and avoiding collision with obstacles, is presented. The genetic algorithm uses a multi-objective function to obtain the waypoints; this functions are the the length of the path, the distance from the waypoints to the obstacles, and the probability of the final trajectory to cross an obstacle within a safe zone. Since a path generated by only the waypoints is discontinuous, these are fed to a continuous path generator to find a trajectory based on parametric equations, considering a minimum radius of turn. Real-time experiments are obtained in order to validate the proposed algorithm.