Doctor of Philosophy
Electrical & Computer Engineering
Indoor environments are a particular challenge for Unmanned Aerial Vehicles (UAVs). Effective navigation through these GPS-denied environments require alternative localization systems, as well as methods of sensing and avoiding obstacles while remaining on-task. Additionally, the relatively small clearances and human presence characteristic of indoor spaces necessitates a higher level of precision and adaptability than is common in traditional UAV flight planning and execution. This research blends the optimization of individual technologies, such as state estimation and environmental sensing, with system integration and high-level operational planning.
The combination of AprilTag visual markers, multi-camera Visual Odometry, and IMU data can be used to create a robust state estimator that describes position, velocity, and rotation of a multicopter within an indoor environment. However these data sources have unique, nonlinear characteristics that should be understood to effectively plan for their usage in an automated environment. The research described herein begins by analyzing the unique characteristics of these data streams in order to create a highly-accurate, fault-tolerant state estimator.
Upon this foundation, the system built, tested, and described herein uses Visual Markers as navigation anchors, visual odometry for motion estimation and control, and then uses depth sensors to maintain an up-to-date map of the UAV's immediate surroundings. It develops and continually refines navigable routes through a novel combination of pre-defined and sensory environmental data. Emphasis is put on the real-world development and testing of the system, through discussion of computational resource management and risk reduction.
© The Author
Is Part Of
VCU University Archives
Is Part Of
VCU Theses and Dissertations
Date of Submission