Intelligent Guidance, Navigation and Control of Multi-Agent UASs with Validation and Verification

Abstract

Following the exponential growth in the usage of unmanned aerial systems (UASs) across the Aerospace Industry, more intelligent and robust guidance, navigation, and control algorithms are vital to cope with increasing levels of mission complexity. Additionally, many unmanned aerial operations require large payloads and long endurance such as extended reconnaissance, large-scale search and rescue and fine resolution terrain mapping. However, the stringent payload of a single agent or small UASs reduces their overall practicality and effectiveness. My research aims to address these inherent limitations of small UASs with a swarm by holding the required formation in order to distribute tasks and payload among multiple UASs. The goal of this research is to overcome the challenges of operating multi-agent systems by developing phasic navigation and guidance algorithms. Aircraft dynamics and their interactions with surrounding agents are highly nonlinear, which makes autonomous formation flight very sensitive to aircraft initial conditions. The phasic navigation algorithms are proposed and consist of hybrid mathematical approaches: Frenet-Serret curvature control, Hungarian algorithm and moving mesh methods. At the first phase, the curvature control allieviates the sensitivity to initial conditions of multi-agent UASs in unstructured environments by matching agents’ heading angle to the united direction. A variation of Hungarian algorithm is implemented with a moving virtual terminal to assign each agent to the formation position. In the second phase of navigation, the moving mesh methods are applied for holding the formation by defining the outer agents’ position for the boundary condition. The significance of the moving mesh methods is a scalability and a inherent intercollision avoidance. Due to the profound difference between the longitudinal and lateral-directional motion of a fixed-wing aircraft, a multi-scale moving point guidance algorithm has been designed to create the separate virtual reference points in the longitudinal and lateral-direction planes for the first time. This method has been shown to greatly reduce tracking oscillations and improve the overall tracking quality and coherency of the formation. Monte Carlo simulations are performed to ensure the stability and robustness of implementing proposed algorithms through an essentially exhaustive search. A broad range of random initial conditions have been used to validate the effectiveness of guidance, navigation, and control algorithms. Another unique contribution of this work is the validation and verification of proposed algorithms by the hardware-in-the-loop testbed and the numerous flight tests. The hardware-in-the-loop testbed is designed to test the avionics and communication before the flight test by simulating onboard 6-degrees of freedom nonlinear equations of motion. Over one hundred flight tests have been conducted using three distinct aircraft platforms between 2016 and 2018 to validate the fundamental building blocks of this architecture. In summary, this dissertation provides a conceptual and practical foundation for guidance, navigation, and control of multi-agent cooperative/collaborative UASs by unique approaches.