Performance Comparison of Integrated Navigation System for Different Configurations

Global positioning systems (GPS) position is determined from the distance measurements to four or more satellites with good geometry. Autonomous vehicles can utilize GPS navigation under open sky areas. However, GPS signal is degraded/denied in urban areas due to satellite signal blockage, multipath, and interference. A suite of multiple types of sensors with distinct strengths is required to achieve localization and positioning for autonomous operation. This paper is a review of a multi-sensor system with different configurations. Two types of the reduced inertial sensor system (RISS) were utilized: Odometer-based RISS/GPS and Frequency Modulated Continuous Wave Radar-based RISS/GPS. The Radar-based RISS/GPS navigation system performance is then evaluated by adding Magnetometer measurements. Fast Orthogonal Search (FOS) nonlinear error modeling is employed for residual non-linear position errors modeling associated with the Magnetometer / Radarbased RISS /GPS positioning solution to improve the results further. During the availability of the GPS, FOS established the non-linear error model. In degraded/denied GPS environments such as urban canyons, the FOS-based non-linear error model operates in prediction mode to estimate the position errors associated with the Magnetometer / Radar-based RISS /GPS, thus providing improved positioning performance. These system configurations were evaluated for an actual road test trajectory for a GPS outage of 300 seconds to examine the performance. The results show that the system configurations play a significant role in improving positioning accuracy.