Модель автоматизированного распределения в задаче группового преследования с защитниками

This article examines a computer model of a quasi-discrete group pursuit game. In which there are pursuers, goals and defenders. In the article’s models, the task of the pursuers is to achieve static goals. Achieving one goal is possible by several pursuers at different times. The task of the defenders is to defeat the pursuers. A win for the pursuers can be considered the achievement of at least one of the pursuers of their goal. The defeat of all targets is considered a win for the defenders. For defenders, the number of pursuers is not certain. In the model of the article, a unified environment for detecting pursuers is formed. A unified detection environment is the result of set-theoretic operations on points in space. In the article’s models, a single detection environment is a simply connected set. In the case of multiply connected sets, model algorithms with different logic would be proposed. The pursuer is considered detected if it enters this area. The assignment of a target defender to the detected pursuer is carried out according to several optimization criteria. The defender may be appointed from the estimated time of achievement. In one implementation of the model, this is the minimum sample time for a given defender. As a variant of the optimization factor, the defender for the pursuer can be selected based on the minimum distance to him. The article also discusses options for localizing defenders at one point. The article proposes four models for consideration. A model of group pursuit with optimization based on the expected time for defenders to reach the pursuers. Group pursuit model with optimization for the initial distance from the moment the pursuers are detected by defenders. Additional algorithms for the above-described models have been developed that take into account the restrictions on the presence of defenders in one location. The difference between algorithms is considered in the case of the presence or absence of information for defenders about the number of pursuers, their goals, etc. The results obtained in the article may be in demand by developers of robotic systems with autonomous control. Ideally, it would look like this: A group of pursuers and a group of defenders, each participant has its own algorithm of actions, would have its own software environment, where each participant performs its task, guided by the rules. That is, this software environment resembles a network real-time strategy game.