Along with advancements in microelectromechanical system (MEMS) technology, many modern personal navigation devices incorporate measurements from various sensors alongside Global Navigation Satellite Systems (GNSS) receivers. Despite using these sensors, GNSS still remains an important component of these navigation devices in view of its absolute positioning capability. Thus, when it comes to navigating in GNSS signal degraded areas like in urban and natural canyons, the performance of such multi-sensor integrated navigation systems is still found to be sub-optimal. In particular, existing filtering algorithms are often unreliable in such environments. This affects the usability of such personal navigation devices in some applications where reliability is a critical parameter. Moreover, reliability can be further degraded by the occurrence of faults in other sensors besides GNSS. This research thus develops several algorithm modules with an ultimate goal of improving the performance, and especially reliability, for low cost multi-sensor integrated navigation systems.
Among the proposed algorithm modules, the first method modifies the filtering algorithm by replacing the assumption of normal distribution of GNSS measurements with that of a heavy-tailed distribution. The second module adapts the covariance of the GNSS measurements to match the true error characteristics of the surrounding environment, based on the consistency of GNSS derived user acceleration values to those obtained from inertial measurement units. Finally, a third algorithm module detects possible faults arising in various sensors. Based on the type of sensor fault, the algorithm either rejects some of the measurements before they enter the integration filter, issues a warning signal to indicate lack of reliability information or deems the navigation solution unusable. The proposed algorithms are tested with numerous field data sets collected in various environments as well as with carefully simulated faults that are added to clean measurements.
The analysis of the results obtained using the proposed methods indicate a significant improvement in the reliability of the navigation solution. The average improvement in the reliability varied between 15 % and 26 % for the data sets used in the analysis. Position accuracy was also found to improve. In particular, maximum position errors are significantly decreased, up to a factor of 2.5 in some cases. Finally, the simulated as well as actual faults occurring in the sensor measurements were also correctly detected.