Modular Heterogeneous Multi-Agent Control Framework with Integrated Payloads

Abstract

Small unmanned aircraft are being used in an increasing number of applications ranging from emergency response to parcel delivery. Many of these applications are benefited when employed as a multiple-vehicle operation. Such operations often require tight cooperation between heterogeneous vehicles and often depend on integration with sensors and payloads. Multi-agent control algorithms can be implemented to control such systems but often require the development of an underlying vehicle communications framework in addition to a sensors and payloads communications framework. This thesis presents a single unified modular framework, named Clark, and supports heterogeneous multi-agent control and sensor/payload integration. Clark provides a wireless network between agents without relying on pre-existing communications infrastructure, and provides software interfaces for connecting to a variety of payloads. This thesis first reviews small unmanned aircraft systems (SUAS), multi-agent control, multi-agent control testbeds, and wireless networking technologies used on SUAS. Systems engineering is then employed to develop an Identified Need, Concept of Operations (ConOps), and requirements. All Defined, Derived, and Design Requirements are explained and justified. Some requirements are highlighted to demonstrate key features of the Clark framework. The software architecture is explained in detail in a top-down approach. Hardware is selected for prototyping and shown to meet the requirements. Bench tests, ground tests, and flight tests are conducted to verify the framework’s ability to communicate between agents and affect control. Ground testing includes a multi-agent cooperative mission while flight testing features two and three agent missions. Test results are presented and demonstrate the candidacy of Clark as a modular heterogeneous multi-agent control framework with integrated payloads.