Advanced navigation architecture for low-cost unmanned aerial vehicles

Mwenegoha, Hery Amani (2022) Advanced navigation architecture for low-cost unmanned aerial vehicles. PhD thesis, University of Nottingham.

[img] PDF (VDM Navigation Hery Mwenegoha PDF Version For Internal Examiner) (Thesis - as examined) - Repository staff only - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
Available under Licence Creative Commons Attribution.
Download (34MB)

Abstract

This thesis details the effective integration of global navigation satellite systems (GNSS) with an inertial navigation system (INS) to meet the requirements for use in small, mass-market unmanned aerial vehicles (UAVs). A key focus is on the addition of the input from the vehicle dynamic model (VDM).

The dominant navigation system for most small, mass-market UAVs is based on INS/GNSS integration. The integration of the two systems provides a navigation solution with both short-term and long-term accuracy. However, during a GNSS outage, the navigation solution drifts. This can happen due to severe multipath, intentional or unintentional interference, even against cryptographically secured GNSS signals, rapid dynamics and loss of line of sight to the satellites. Most small UAVs use low-cost inertial sensors, which during a GNSS outage, will cause the navigation solution to drift rapidly. Traditionally, additional aiding sensors such as cameras and range finders have been used to reduce the rapid drift of the navigation solution. However, this approach adds extra weight and additional cost to the overall system. More recently, the use of a VDM in providing improved navigation performance has gained research popularity, especially for small, mass-market UAVs. This approach preserves the autonomy of the navigation system while avoiding extra cost, additional weight, and power requirements, essential for low-cost applications.

This thesis presents a VDM navigation architecture suitable for a fixed-wing UAV fitted with low-cost inertial sensors and a GNSS receiver during periods of extended GNSS outage. The thesis presents and examines state-of-the-art VDM navigation techniques, quantifies their limitations and identifies approaches to reduce navigation solution drift during GNSS outages. An integration algorithm that implements the approaches and overcomes these limitations is developed and evaluated via a Monte Carlo simulation study. The integration algorithm is then tested on real flight data gathered from a test flight using a small fixed-wing UAV.

The thesis identifies that most current VDM integration schemes use a loosely coupled configuration, using position and velocity measurements from a GNSS receiver. This work shows that the VDM navigation solution can drift significantly with this configuration during an extended GNSS outage. A novel VDM-based architecture is then developed to reduce the navigation solution drift during extended GNSS outages. The architecture, referred to as a tightly coupled VDM-based integration architecture (or simply TCVDM), uses raw GNSS observables and measurements from inertial sensors to aid the navigation solution even when tracking less than four satellites. The architecture uses an extended Kalman filter (EKF) to estimate the navigation solution errors. A software-based GNSS measurement simulator is also developed to generate the raw GNSS observables.

Simulation results reveal significant improvements in navigation accuracy during GNSS outages. In the case of a GNSS outage lasting over two minutes, results show that position accuracy improves by one to two orders of magnitude compared to a tightly coupled INS/GNSS integration scheme (TCINS) and by a factor of four compared to the state-of-the-art VDM integration architecture. In addition to the navigation states, the filter also estimates wind velocity components, VDM parameters and the receiver clock offset and drift. The estimation of wind velocity components is achieved even without an air data system. It is found that the architecture only resolves 40% of the initial error in the model parameters. This is found to be sufficient for navigation with randomly distributed errors of 10% in the model parameters.

The developed architecture is also tested on real flight data gathered using a small fixed-wing UAV. A custom flight control system (FCS) that houses a low-cost inertial measurement unit (IMU), barometer and a data logging module is used on the UAV. The FCS is used for guidance, navigation and logging control inputs and different measurements. Three GNSS receivers are installed on the UAV and used to derive a reference position, velocity and attitude solution. Test results show that the position error estimation performance for the TCVDM scheme improves by a factor of 43 compared to a TCINS scheme with two satellites visible during a GNSS outage. The velocity error estimation performance for the TCVDM scheme improves by a factor of 7 across all channels compared to a TCINS scheme during the outage. However, the TCVDM scheme shows poor attitude estimation performance. This is attributed to the lack of accurate VDM parameters, especially the torque coefficients, which leads to significantly worse yaw angle estimation performance.

This work presents an alternative, low-cost navigation scheme for small UAVs that uses sensors usually available in most UAVs. The navigation scheme can work alongside existing integration architectures to provide a secondary navigation solution for improved reliability and integrity monitoring.

Item Type: Thesis (University of Nottingham only) (PhD)
Supervisors: Pinchin, James
Jabbal, Mark
Keywords: Vehicle Dynamic Model, VDM, Tightly Coupled INS/GNSS, Tightly Coupled VDM, Unmanned Aerial Vehicles, UAVs
Subjects: T Technology > TL Motor vehicles. Aeronautics. Astronautics
Faculties/Schools: UK Campuses > Faculty of Engineering
Item ID: 67196
Depositing User: Mwenegoha, Hery
Date Deposited: 31 Jul 2022 04:40
Last Modified: 31 Jul 2022 04:40
URI: http://eprints.nottingham.ac.uk/id/eprint/67196

Actions (Archive Staff Only)

Edit View Edit View