Underwater Rotation MEMS-IMU/DVL Integrated Navigation Based on IFB-FGO

Abstract

The rotation MEMS inertial measurement unit (IMU)/Doppler velocity log navigation system has the potential to outperform expensive fiber-optic gyroscope systems in underwater navigation. Factor graph optimization (FGO) can refine all historical states, which fits the requirements of this navigation system. Traditional FGO methods, however, demand exact initial attitudes, and when accounting for the earth's rotation, IMU preintegration requires frequent state-related recalculations, leading to high computational costs. To address this, we propose inertia-frame-based factor graph optimization (IFB-FGO), which obviates the need for initial attitude information and accelerates computations. By isolating the initial attitude at the start time as a separate state and constraining the initial misalignment angle to zero, we enhance the linearity of the error dynamics system, thus improving stability and precision. In addition, our design for IMU preintegration is decoupled from the state, thus requiring only a single calculation and substantially reducing computational costs. Experimental findings confirm that IFB-FGO outperforms existing methods with reduced computational costs.