A Fast MEMS-IMU/GPS In-Motion Alignment Method Using Full-Integration-Based Position Loci

The initial alignment is essential to guarantee the navigation performance of a strapdown inertial navigation system (SINS) and global positioning system (GPS) integration. Since a position reference vector is usually taken as an output, current filtering-based in-motion alignment methods that use position loci can only employ partial integration to reduce the accumulative inertial measurement unit (IMU) bias errors. However, the partial integration can amplify the negative impact of the GPS outliers, resulting in degraded alignment performance. In order to address this issue, we take the position reference vector as an augmented state to suppress the accumulative IMU bias errors, so that the full-integration-based position loci can be applied to reduce the negative impact of the GPS outliers. Specifically, we use an on-manifold error state extended Kalman filter (ESEKF) with the proposed full-integration-based state-space model to estimate position reference, velocity reference, attitude error, and IMU bias vectors. Moreover, since the widely used interpolation procedure can be avoided in our framework due to the existence of the full-integration-based state, we do not need to update the state at each IMU sampling instant, and thus the computational burden is greatly reduced. Experimental results show that our method achieves higher alignment precision than existing methods using position loci while maintaining a similar running time.