Novel Integration Strategy for GNSS-Aided Inertial Integrated Navigation

Abstract

The conventional integration mechanism in GNSS (Global Navigation Satellite Systems) aided inertial integrated positioning and navigation system is mainly based on the continuous outputs of the navigation mechanization, the associated error models for navigation parameters, the biases of the inertial measurement units (IMU), and the error measurements. Its strong dependence on the a priori error characteristics of inertial sensors may suffer with the low-cost IMUs, e.g. the MEMS IMUs due to their low and unstable performance. This paper strives for a significant breakthrough in a compact and general integration strategy which restructures the Kalman filter by deploying a system model on the basis of 3D kinematics of a rigid body and performing measurement update via all sensor data inclusive of the IMU measurements. This novel IMU/GNSS Kalman filter directly estimates navigational parameters instead of the error states. It enables the direct use of the IMU's raw outputs as measurements in measurement updates of Kalman filter instead of involving the free inertial navigation calculation through the conventional integration mechanism. This realization makes all of the sensors in a system no longer to be differentiated between core and aiding sensors. The proposed integration strategy can greatly enhance the sustainability of low-cost navigation systems in poor GNSS and/or GNSS denied environment compared to the conventional aided error-state-based inertial navigation integration mechanism. The post-processed solutions are presented to show the success of the proposed multisensor integrated navigation strategy.