IMFlySim: A High-Fidelity Simulation Platform for UAV Swarms

The development of unmanned aerial vehicles (UAVs) shows the development trend of high intelligence and UAV Swarms, which makes the UAV Swarms flight trails in the real environment expensive and time-consuming. The current simulation platforms that can serve UAV swarms have problems such as insufficient number of nodes, low visual rendering fidelity, lack of intelligent interfaces, and poor real environment reproduction. This paper proposes a high-fidelity simulation platform for UAV swarms, which is a multi-node simulation platform based on Client/Server (C/S) architecture with Unreal Engine as the rendering engine and JSBSim as the physical engine. JSBSim can realize multi-model, high-frequency hardware-in-the-loop data simulation and Unreal Engine provides data interface for intelligent sensing and decision algorithms. The C/S architecture can increase the maximum number of hardware-in-the-loop simulation nodes to meet the UAV swarms flight requirements. In this paper, simulation experiments of fixed-wing are conducted and compared with the flight test, the error is between -2.851017 and 3.108375, with a mean of -0.034536 and a variance of 0.476047. The error always fluctuates around 0, and this result is not bad. Finally, this paper successfully realized formation flight simulation.