An Improved Genetic Algorithm for Cooperative Allocation with Orbit-Like Tasks

This paper addresses the problem of cooperative allocation with the optimization of completion time, energy cost and at the same time obstacle avoidance, where tasks are orbit-like, that is, simple, convex closed orbits. Some are executed by one agent while the others are executed by multiple agents. A lagrange interpolation simplification algorithm (LISA) is designed to extract limited feasible points from each orbit, which makes possible to use
$A$
* algorithm to calculate the distance matrix for the genetic design. A so-called node selecting genetic algorithm (NSGA) is designed to calculate the optimal node selection scheme for the optimal node selection, which is integrated into the so-called task allocating genetic algorithm (TAGA) to achieve the optimization of cooperative allocation. To accelerate the convergence of the TAGA, the parent of crossover operation selects the optimal individual, the optimal group and its own reverse order according to the way of probability selection and the initialization operation starts by placing an agent at a short distance from the work area. The effectiveness of the algorithm is verified by a simulation example of multi UAV cruise, which indicates that the average number of stable iterations is 58.7% less as compared with the ordinary genetic algorithm.