A comparison of loosely-coupled mode and tightly-coupled mode for INS/VMS

Abstract

Misalignment between the inertial navigation system (INS) body axes and vehicle body frame (VBF) can incur serious errors in the integration of INS and a mechanical odometer of vehicle motion sensor (VMS) on vehicles. In this contribution, two INS/VMS integration architectures are discussed, including loosely coupled mode and tightly coupled mode. The aim is to offer solutions to the problem of INS-to-VBF alignment and figure out how the INS-to-VBF misalignment affects different integration architectures. In the first architecture, the VMS outputs and nonholonomic constraints are used to calculate the velocities and positions, leaving the INS with the role of determining the vehicle attitude. In such integration, INS-to-VBF alignment and calibration of VMS scale factor are implemented by a geometrical method in advance. In the second architecture, the INS and VMS are tightly integrated in a single Kalman filter to form corrections to the INS error states and corrections to the VMS error states. The advantages and shortcomings of these two architectures are comprehensively analyzed. Ground based experimental results show that INS-to-VBF misalignment leads to more errors in loosely coupled integration than tightly coupled integration.