UAV Autonomous Navigation by Data Fusion and FPGA

Abstract

Currently, the use of unmanned aerial vehicles (UAV), also known as drones, is increasing. The applications are in several areas such as engineering projects, agriculture, livestock, monitoring, and rescue. One of the main reasons to use UAV is its lower cost when compared to manned aircraft. The flight of a UAV can be done remotely or autonomously. For the autonomous navigation, a Global Navigation Satellite System (GNSS) is usually applied. However, a GNSS system can suffer natural or human interference, becoming the research for alternatives strategies a hot topic in this field. An approach to carry out the autonomous navigation without use of GNSS signal is to estimate the UAV position by using data fusion combining different sensors. A solution for autonomous navigation is presented applying inertial sensor and image processing, both are employed to estimate the drone position. The data fusion process is carried out by a computational intelligence procedure. Two self-configuring ANNs are employed here: for image edge extraction, and an operator for data fusion. A hybrid computer architecture is employed to implement the solution with standard CPU and FPGA (Field Programmable Gate Array).