A multi-sensor data fusion navigation system for an unmanned surface vehicle

Abstract

Worldwide there is an increasing interest in the development of unmanned surface vehicles (USVs). In order for such vehicles to undertake missions, they require accurate, robust, and reliable navigation systems. This paper describes the implementation of a fault tolerant autonomous navigation approach for a USV named Springer. An intelligent multi-sensor data fusion navigation algorithm is proposed that is based on a modified form of a federated Kalman filter (FKF) utilizing a fuzzy logic adaptive technique. The fuzzy adaptive technique is used to adjust the measurement noise covariance matrix R to fit the actual statistics of the noise profile present in the incoming sensor measured data using a covariance matching method. Information feedback factors employed in the FKF are tuned on the basis of the accuracy of each sensor. In order to compare the fault-tolerant performance, several fuzzy-logic-based cascaded Kalman filter architectures are also considered. Simulation results demonstrate the algorithm's capability under different types of sensor fault.