A Simulation Software Architecture Design Suitable for Multi-Agent Aircraft

With the surge in demand for collaborative control, intelligent perception and independent decision-making of drones clusters, multi-agent aircraft simulation technology has become the core supporting tool to promote the development of intelligent aviation systems. The design of the simulation software architecture directly determines the efficiency of collaborative training of large-scale agents and the reliability of virtual and real migration. This paper proposes a new simulation software architecture in response to the problems of insufficient scalability of parallel simulation structures of the current multi-agent simulation architecture, real-time nature of high-speed data distribution, and limitations of simulation acceleration technology. Through distributed task scheduling engine, layered data bus and heterogeneous computing resource pooling technology, efficient simulation and strategy migration verification of large-scale aircraft clusters are achieved, providing reliable support for the actual deployment of intelligent aviation systems, and developing a simulation platform to verify the simulation efficiency of multiagent aircraft.