Low Cost, High Integrity, Aided Inertial Navigation Systems for Autonomous Land Vehicles

Abstract

This thesis describes the theoretical and practical development of a low cost, high integrity, aided inertial navigation system for use in autonomous land vehicle applications. The demand for fail safe navigation systems which can be used on large autonomous land vehicles such as those found in container terminals, agriculture, construction and in mines, has driven research and technology into the development of high integrity navigation suites. Integrity, in this thesis, is defined as the ability of a navigation system to provide reliable navigation information while also monitoring the health of the data and either correcting any faults that may occur or rejecting faulty data. Thus integrity encapsulates reliability while the reverse is not necessarily true. This thesis provides, both in practical and theoretical terms, the fusion processes adopted and the implementation of fault detection techniques required for a high integrity aided inertial navigation system. There are three main contributions: Firstly, the development of an aided inertial navigation system using the Global Navigation Satellite System (GNSS) as an aiding source for use in autonomous land vehicles. This is accomplished by using a Kalman filter as the estimation algorithm along with the addition of fault detection techniques so as to increase the integrity of the system. The real time structure of the navigation architecture is provided along with results of its implementation in an autonomous 65 tonne straddle carrier. However, the algorithm development provides a generic structure thus allowing the use of the navigation suite on any land vehicle. Secondly is the use of vehicle modelling to bound drift errors associated with inertial navigation. This provides a sensor-free aiding source due to the inherent constrained motion of land vehicles. Vehicle constraints can thus be used as an extra aiding source with other sensors which in turn improves the accuracy and integrity of the overall navigation system. This is demonstrated with the real time implementation of an inertial navigation system being aided by three separate aiding sources; GNSS, vehicle modelling and speed data provided by an encoder. Finally, the understanding of the effect of inertial sensor redundancy to navigation accuracy and fault detection is addressed. A redundant inertial measurement unit is developed and tested and provides the necessary physical sensor for future fault detection work. This concludes this thesis by providing the foundation for the autonomous detection of faults in inertial units and furthermore the final level of integrity required for a navigation system.