Sensor Failure Scenarios and Corresponding Solutions of Three Axis Magnetometer Integrated GPS/INS UAV Navigation System

Abstract

In this thesis, measurement of 3-axis magnetometer is integrated with GPS/INS navigation system in order to improve the navigation performance and to solve the problem of performance degradation caused by sensor failure. GPS/INS integrated navigation system is widely used as a small UAV navigation system. GPS/INS integrated navigation system has excellent navigation performance by combining stable navigation performance of GPS with high accuracy navigation performance of INS during a short time. However, the GPS/INS integrated navigation system cannot observe the attitude error in the static condition like horizontal straight flight, and it is impossible to observe the attitude error completely even when the maneuver such as linear acceleration or horizontal turning is performed. In order to compensate for this lack of observability, magnetometer is additionally integrated to the geomagnetic sensor. There are two methods of magnetometer integration: a magnetic heading integration method and magnetic vector integration method. The magnetic heading integration method determines the heading from the magnetic field measurement, assuming that the roll angle and the pitch angle are known, and then integrates the heading information with the navigation system. On the other hand, the magnetic vector integration method doesnt require the information about the roll angle and the pitch angle, and the attitude information of the two axes perpendicular to the magnetic vector is provided. However, the use of the magnetic vector integration method with inaccurate magnetic measurements may adversely affect navigation performance. Therefore, mainly the magnetic heading integration method have been widely used. The factors that cause errors in the measurements of the magnetometer are the current flowing around the sensor and the influence of the metal located around the sensor. Small UAV mainly uses electric motors to drive the propeller. Therefore, electric currents that drive the motors to the electric wires connected between the battery and the electric motors cause errors in the magnetometer located around the electric wires. When the current sensor is installed and the measured current values are used, the magnetic field error of the bias type generated by the current can be compensated, but there is a problem that the noise of the magnetometer measurement value is increased due to the PWM driving current characteristic of the motor. In this thesis, a compensation method of the magnetic field error by combining the throttle input and the current measurement by deriving the relation between the throttle input and the motor drive current is proposed. This method is effectively compensating the measurement error of the bias and noise type geomagnetic sensor. A vector calibration method based on flight data is proposed as a method to compensate for magnetic field distortion caused by metal objects. This method precisely models the magnetic field distortion caused by the surrounding metal using the attitude information of the GPS/INS integrated navigation and the spherical harmonics model (SHM) and has superior performance compared to the scalar based calibration method such as the existing ellipsoid calibration method. In this way, magnetometer calibration methods suitable for small UAV is proposed, and these methods improve the accuracy and usability of the magnetometer measurements. In this thesis, the magnetic vector integration is performed in the GPS/INS navigation system using the high accuracy magnetic field measurements through the proposed magnetometer calibration method. When GPS is available, the accuracy of the attitude is improved compared with the conventional magnetic heading integration method. In the case where GPS is not available, stability of the navigation system is improved by reducing divergence of velocity and position error as well as attitude error. On the other hand, GPS/INS integrated navigation cannot be performed in case of IMU failure. Position and velocity information can be obtained from GPS receiver, but alternative systems for attitude estimation are required. In this thesis, attitude error of the single-antenna GPS attitude determination method is estimated by integrating 3-axis magnetometer. In addition, the performance of the alternative attitude estimation system is improved by subdividing the fault situation of the IMU and integrating the available measurements in each situation. In this thesis, the failures of various sensors constituting a small UAV GPS/INS navigation system are assumed and the stability and survivability of the UAV are improved by reducing the performance degradation of the navigation system in each situation