Implementation Of An Autonomous Navigation Stack For UAVs In Tightly Closed Industrial Spaces
Master thesis
Permanent lenke
https://hdl.handle.net/11250/3020377Utgivelsesdato
2022Metadata
Vis full innførselSamlinger
Beskrivelse
Full text not available
Sammendrag
Automation of industrial infrastructure inspection and surveillance became an integratedpart of the new industrial environments in the age of the fourth industrial revolution anddigital transformation; thus, the need to remove manual labor from such processes to im-prove efficiency and reduce costs became necessary. Recent advances in sensor technology,edge devices, and Unnmaned Aerial Vehcile (UAV) control systems have made it possible todevelop systems with such capabilities using off-the-shelf components and integration plat-forms.This thesis project aims to identify and develop a navigation stack for localization and nav-igation in the absence of Global Positioning System (GPS) signals. Furthermore, industrialenvironments are frequently complex in terms of navigation space; thus, such systems musthave a high precision rate for pose estimation. Tanks and pipes for storage and transport,which require inspections for safety and are usually costly and risky, are also common inan industrial environment. As a result, UAVs can offer a profitable inspection solution instructures that offer precision and speed.In this work, a working prototype is developed for a system that is capable of navigating aUAV in a simulated industrial environment using a local path planning module with esti-mated pose data from a localization module using visual odometry sensor data. The resultsshow an average error rate of 10-30 cm, which is promising in navigation within complexspaces with a range of 1.5-8.5 meters.The developed stack can serve as the foundation for future development to improve poseestimation precision and performance while also increasing the stability of the local pathplanning algorithm. Furthermore, the developed system represents a one-way communicationchannel between the localization module and the local path planning module. However, bi-directional communication to provide feedback from the local path planning module to thelocalization module can improve the precision.iv