One special problem of UAV routing

• Main Authors: Л. Ф. Гуляницький, В. В. Сторчевий
• Format: Article
• Language: English
• Published: State University “Uzhhorod National University” 2019-07-01
• Series: Науковий вісник Ужгородського університету. Серія: Математика і інформатика
• Subjects: безпiлотнi лiтальнi апарати; оптимiзацiя маршруту; мультидепо; математична модель; комбiнаторна оптимiзацiя; мурашина система
• Online Access: http://visnyk-math.uzhnu.edu.ua/article/view/180954

The purpose of the article is to review the tasks of routing vehicles, their classification, as well as consideration of the problem of routing the group of unmanned aerial vehicles (UAVs) during the survey of specified objects. The proposed mathematical model of the problems is that before a given group of aircraft, which can start from different points of launch and must complete the route in special receiving zones having a certain capacity, the task is to fly a number of specified objects (points on the ground) minimizing the total length of the routes or the duration of the flights, provided that each object is visited by one and only one UAV and all objects must be visited. It is shown that the model of traffic of vehicles with multiple depots can be adapted for flight planning of the UAV group in a number of survey tasks. The next solution algorithms are offered: brute force, hybrid local search - tabulated search using greedy algorithm for constructing an initial route, max-min algorithm of ant system (MMAS). Since there are no data libraries for formulated task statements in open access, a set of tasks was generated using the Google Maps API, which allowed you to get tasks that are close to reality (using the coordinates of villages, towns, cities). Tasks have a different number of surveyed objects, aircraft, start and finish depots. For each algorithm, a selection of parameters was made and then a computational experiment was conducted, which allowed comparing the three developed algorithms for solving the routing problem of the UAV group. As a result, it is determined that the method of brute force can be applied only to problems of very small dimension, while it takes a lot of time to execute the algorithm. The local search found improved solutions to problems of medium dimension, and the max-min algorithm of ant system proved to be the best on problems of large dimension, therefore it is defined as perspective for research. The further direction of improvement of the max-min algorithm of ant system is determined by using the principles of search diversification in them.