COMPUTATIONALLY EFFICIENT ALGORITHMS FOR OPTIMAL MOTION PLANNING AGAINST MULTI-DOMAIN SUPER SWARMS

This thesis develops robust tactics for countering multi-domain super swarms. Previous research has provided tools for assessing adversarial swarms’ internal cooperating strategies, quantifying risk based on swarm and weapons models, and generating optimal defender trajectories. In this research, we develop a simulation-based testbed for experimental validation of these strategies and a database of adversarial swarming models against which to test. In this research, the aforementioned simulation-based testbed is examined from the perspective of computational efficiency. A significant computational advantage is obtained by replacing the Runge Kutta with the Verlet integration scheme frequently used in the molecular dynamics community. This almost equally numerically stable framework offers an impressive performance advantage. Additionally, this research seeks to find the ideal balance between the deterministic nature of the dynamics of adversarial swarms and the requirement for a probabilistic approach in order to model the mutual attrition between opposing agents during combat situations. To achieve this end, several models of dynamics and attrition are introduced for optimal motion planning. Their outcomes are compared with a Monte Carlo simulation model, in which survivability is partially influenced by random number generation that aims to simulate the unpredictability exhibited in the real world.Outstanding ThesisLieutenant, Hellenic NavyApproved for public release. distribution is unlimite