An immune plasma algorithm with Q-learning based pandemic management for path planning of unmanned aerial vehicles

Abstract

The countries have experienced the tremendous potential of unmanned aerial vehicles and their military counterparts in recent years. For further improving the task performances of these autonomous vehicles, their flight paths should be determined or calculated optimally by taking into account enemy weapon systems, fuel or battery usage and some limitations about the turning, climbing or diving angles. Immune Plasma algorithm (IP algorithm or IPA) is the first intelligent optimization technique modeling the details of an infection treatment method called convalescent or immune plasma gained popularity again with the coronavirus disease and showed its promising performance for various engineering problems. In this study, Q-learning that is a reinforcement learning algorithm was integrated into the workflow of the IPA for managing some pandemic measures including lockdown, partial opening and full opening. Moreover, the treatment schema was completely changed in order to improve the search efficiency and remove the requirement of algorithm specific control parameters. The newly introduced IPA variant also named Q-learning IPA (Q-LIPA) was tested with the purpose of planning paths and a set of detailed experiments was carried out over twelve test cases of three different battlefield scenarios. The paths found by Q-LIPA were compared with the paths of well-known intelligent optimization techniques and their modifications. Comparative studies indicated that both Q-learning based pandemic measure management and specialized treatment schema positively contribute to the solving performance and help Q-LIPA to outperform its rivals for the majority of the test cases.