Navigation of UAV in Denied GNSS Environments Using Multi-Sensor Systems
Date
2018-08-10
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
There have been extensive market demands over the past 10 years for deploying small autonomous Unmanned Aerial Vehicles (UAVs) in enormous civil and military applications such as search and rescue, disaster management, firefighting, reconnaissance and border mentoring. While UAVs are performing their missions, they are typically relying on the onboard Global Navigation Satellite System (GNSS)/ Inertial Navigation System (INS) integrated navigation system for the positioning and localization purpose. During such missions, the GNSS signals could be prone to blockage, attenuation, multipath effect, jamming and spoofing. In such complicated scenarios, the navigation solution is acquired by the INS in standalone mode prior to the GNSS signals recovery. Consequently, the navigation solution will deteriorate rapidly because of the drift exhibited by the low-cost INS during GNSS signal outages. Therefore, the necessity for an accurate and reliable navigation system in such cluttered environments is essential to achieve their missions. A variety of sensors and techniques have been exploited in an attempt to provide a reliable navigation solution in GNSS-denied environments. Although these sensors have some strengths individually, they still suffer from some limitations. Monocular Visual Odometry (VO) has been proposed as a GNSS denied environment navigation system for UAVs since it has light weight, small size and low power consumption. This monocular VO suffer from the scale ambiguity if there is no other aiding sensor or prior information of the observed scene. Furthermore, it depends on a rigorous calibrated camera and system model which may change from one flight to another or even during the flight. Therefore, a novel monocular VO based on optical flow and regression tress is proposed which eliminates the need for a calibration phase and inherently models the interior camera parameters, its lever arm and boresight parameters since, the relationship between the actual optical flow vectors and the navigation states are implicitly modeled during the flight. In addition, this monocular VO can resolve the scale ambiguity problem by implicitly modeling the scale on its trained regression model. Although this monocular VO has such capabilities and benefits, its 3D positioning accuracy is still affected by some factors such as the lack of the observed features, inconsistent matches, and the accumulated positioning drift errors. Hence, a smart hybrid vision aided inertial navigation system (VAINS) is proposed to correct both monocular VO and INS drift errors based on trained Gaussian Process Regression (GPR) against GNSS reference data. Although a variety of VO based approaches have been proposed to enhance the navigation solution during the GNSS signal outage, their imagery measurements are affected by brightness, lighting conditions and featureless areas. In addition, their measurements are not immune against the environmental conditions such as rain, fog and dust which could affect their usage as a GNSS denied environment navigation system. In order to avoid such limitations, a lightweight Frequency Modulated Continuous Wave (FMCW) Radar Odometry (RO) aided navigation system is proposed as a GNSS denied environment navigation system for UAVs. This system is immune to these environmental changes and it has light weight, small size, and low power consumption which make it more appealing to be mounted on small UAVs. Although the camera has some strengths and limitations, its incorporation with radar will enhance the performance and will provide a more reliable navigation solution. In addition, the scale ambiguity of the monocular VO is resolved by the estimated RO height. Furthermore, this integrated system is more robust against the environmental conditions since the radar is immune against these environmental changes.
Description
Keywords
Citation
Mostafa, M. M. A. (2018). Navigation of UAV in Denied GNSS Environments Using Multi-Sensor Systems (Doctoral thesis, University of Calgary, Calgary, Canada). Retrieved from https://prism.ucalgary.ca. doi:10.11575/PRISM/32809