Research on inertial/visual navigation with heading-position constraints and the observability analysis

Abstract

Accurate heading reference and familiar landmarks are the key information to guide migratory birds to realize long-distance migration, which provides an important reference for the navigation of unmanned platforms. This paper focuses on the integrated inertial/visual navigation with heading constraints of magnetic compass and position constraints of place recognition. The model of the integrated navigation system is constructed based on extended Kalman filter (EKF), with the derivation of state and observation equations. The observability of system states under different constraints is analyzed based on singular value decomposition (SVD) method. It shows all states of the system are either completely observable or the errors are bounded by introducing the heading and position constraints, which can guarantee the requirements of long-endurance, long-distance and high-precision navigation.