A distributed embedded software architecture for multiple Unmanned Aerial Vehicles

Abstract

In order to deploy intelligent, next-generation applications on Unmanned Aerial Vehicles (UAVs), we must first develop a software architecture that supports onboard computation and flexible communication. This thesis presents an Onboard Planning Module developed from an embedded PC/104 Linux-based computer that communicates directly with the UAV's autopilot to retrieve telemetry data and update the UAV's flight path. A serial communication program exchanges data with the UAV's autopilot while a multithreaded module enables concurrent onboard Mixed-Integer Linear Programming (MILP) optimization. The Mission Manager Graphical User Interface (GUI) monitors the status of each Onboard Planning Module on a team of UAVs using the onboard planning protocol. Two task assignment scenarios are simulated to demonstrate the system operating with both a single and multiple UAV task selection algorithm.