The Simulation of Mosaic Air Combat Based on Agent and Behavior Tree

The U.S. Air Force has proposed the concept of mosaic combat, which combines manned aircraft with low-cost unmanned aircraft to build a mosaic formation. When some systems are damaged, they can be quickly replaced to enhance the reliability of manned and unmanned combat systems. The U.S. military continues to advance research projects such as Skyborg and ACE, but due to the lack of public information, it is difficult to know their true combat modes. This paper aims to achieve mosaic combat simulation. By Reference on the design principles of artificial intelligence built into the U.S. military AFSIM combat simulation software, and using intelligent agents and behavior tree technology, a multi-level mosaic combat decision model is initially constructed. The model is divided into two layers. The upper model mainly realizes the command and control function to complete the target and task allocation, and the lower model realizes the platform's coordinated tactical flight and weapon attack. The combat simulation under typical combat scenarios was completed, and the effectiveness of the combat simulation model based on agents and behavior trees was verified. Finally, the typical factors affecting the victory of mosaic air combat were analyzed through large sample simulations.