A Generic Inertial Navigation System

Abstract

In order to control the movement of a vehicle, precise position and movement information is required. This information is usually obtained through a data fusion of different sensors. This paper presents a new approach of a generic fusion filter for integrated navigation systems, which can be integrated for different types of vehicles. In addition, the fusion filter offers the possibility to add additional sensors with minimal development effort to improve the accuracy of the estimation. For a generic approach a quaternion-based multi-sensor Square-Root Spherical Simplex Unscented Kalman Filter is used. In this way, the nonlinear process and measurement models can be directly applied to the state estimation without the need for analytic derivations or Jacobians compared to conventional Extended Kalman Filter approaches. The presented method was evaluated by sea trials using an unmanned surface vehicle.