Autonomous And Cooperative Multi-UAV Guidance In Adversarial Environment
MetadataShow full item record
The research presented in this dissertation is aimed at developing rule-based autonomous and cooperative guidance strategies for UAVs to perform missions such as path planning, target tracking and rendezvous while reducing their risk/threat exposure level, and avoiding threats and/or obstacles by utilizing measurement information provided by sensors. First, a mathematical formulation is developed to represent the area of operation that contains various types of threats, obstacles, restricted areas, in a single framework. Once constructed, there will be no need to distinguish between adversaries as the framework already contains the information on what needs to be avoided and the level of penalty for a given position in the area. This framework provides the mathematical foundation for the guidance strategies to make intelligent decisions during the execution of the mission and also provides scalar metrics to assess the performance of a guidance strategy in a given mission. The autonomous guidance strategies are developed by using a rule-based expert system approach with the requirements of completing assigned mission or task, avoiding obstacle/restricted-areas, minimizing threat exposure level, considering the dynamic and communication constraints of the UAVs and avoiding collision. All these requirements and objectives are quantified and prioritized to facilitate the development of guidance algorithms that can be executed in real--time. Cooperation of UAVs is modeled by minimizing a cost function, which is constructed based on the level of threat exposure for each UAV and distance of each UAV relative to the target. This improves the performance of the system in the terms of increasing the total area of coverage of the sensors onboard the UAVs, increasing the flexibility of the UAVs to search for better trajectories in terms of obstacle avoidance and threat exposure minimization, and improving the estimation by providing additional measurement. Finally, the performances of the algorithms are evaluated in a simulation environment, which includes the dynamics of each vehicle involved, the models of sensor measurement and data communication with different sampling rates, and the discrete execution of the algorithms. The simulation results demonstrate that the proposed algorithms successfully generates the trajectories that satisfy the given mission objectives and requirements.