ARCHITECTURE OF A SIMULATION ENVIRONMENT FOR REAL-TIME MULTI-AGENT SYSTEMS IN THE TASK OF TEAM OPPOSITION

Panteleyev Michail G., Lityagin Semyon M.

This paper presents the architecture of a simulation environment for real-time multi-agent systems (RT MAS SE) designed for testing scenarios of team opposition. Based on the identified requirements for RT MAS SE and the analysis of existing solutions, a client-server architecture is proposed that ensures scalability across distributed computing resources and modularity of software components. The server part of RT MAS SE implements the modeling of environment dynamics and sensor data, which are transmitted to clients containing intelligent agents and implementing their logic (including team interaction strategies). The paper describes the main interface modules that allow developers to implement various physical platforms, their motion models, sensors and sensor data, as well as to organize network communication using different technologies without changing the overall logic of the RT MAS SE. The limitations of the proposed architecture are identified: the absence of terrain modeling and collision detection between physical platforms; unified storage of sensor data for different sensors in a single container; simplified communication logic between agents. The results of this study can be applied to the development and testing of real-time multi-agent systems in tasks of team opposition. Future work includes extending the architecture by modeling terrain maps and support for training reinforcement learning (RL) agents.

Keywords:team opposition, distributed computing resources, client-server architecture, intelligent agent, simulation environment, real-time multi-agent system.