Connectivity, Resource Integration, and High Performance Reconfigurable Computing for Autonomous UAVs

In an investigation into the capabilities of small autonomous formations of unmanned aerial vehicles (UAVs), we identified connectivity, processing power, and lack of resource integration as three major limiting factors of current technology. In an endeavor to address these issues, we propose a new novel hardware and software environment consisting of a traditional Von Neumann processor coupled with a field programmable gate array (FPGA) for high performance processing, along with support libraries to better manage the resources of a formation. The supporting software libraries have the primary functions of allowing any networked resource (such as processors and UAV sensors) to be accessed from any location in the UAV formation, and also provide support that allows algorithms implemented simultaneously on the reconfigurable and traditional processors to migrate between UAVs for better connectivity to resources or to balance processing loads. In this paper we present the issues we faced in the design of these systems, along with our preliminary results indicating the advantages and shortcomings of the system. We also describe in detail the construction of the prototype systems used to determine the correct software settings for the mobile algorithms