Overview of architecture for GPS-INS integration

Abstract

GPS and INS are two important sensors for providing position and attitude information for geographical referencing systems. The GPS signal alone, blocked by buildings and mountains, cannot give continuous and reliable position all the time and requires high GDOP. Similarly, INS signals, though not affected by surroundings, deteriorate with time. The limitations of GPS and INS can be overcome by integration of these two systems as their characteristics are complementary, and hence a combined system can provide greater autonomy and accuracy. The short time accuracy and high availability of INS combines well with long term accuracy of GPS to provide a more robust and reliable outcome than each of the stand-alone systems. Though Kalman filter and its variants are the most popular approaches, these approaches make certain assumptions, and hence have limitations. This survey paper gives an insight into the various architectures and approaches that were adopted for integration of GPS and INS. Various sensor (data) fusion methods like generic particle filters, DCM and AI techniques like fuzzy logic, expert systems and neural networks were also used. However, all of them have limitations under various conditions, and hence search for better techniques is still on. Though GPS is being referred to in this paper, the concept is equally applicable to all other types of GNSS signals like GLONASS, Galelio, BeiDou, QZSS etc.