Reliable integrated navigation system based on adaptive fuzzy federated Kalman filter for automated vehicles

Abstract

There is an increasing interest worldwide in developing automated vehicles on the highway. From the viewpoint of safe driving, their navigation system should be accurate, robust, and reliable. This paper presents a fault-tolerant navigation approach for automated vehicles based on multi-sensor integration utilizing a proposed adaptive fuzzy federated Kalman filter (AF-FKF). The FKF model is first discussed in detail, which fuses multiple and redundant sensors incorporating strapdown inertial navigation system, carrier phase-differential global positioning system, electronic compass, machine vision, and digital map. The adaptive fuzzy FKF algorithm is then proposed to adjust the FKF information-sharing factors adaptively, detect, and isolate the faulty sensor in the FKF effectively. In order to compare the fault-tolerant performance, several traditional navigation methods are also considered. Simulation results demonstrate that the proposed integrated navigation approach can detect, isolate, and accommodate different types of sensor failures including hard failures and soft failures. The proposed approach can adapt to highly reliable navigation requirements for automated vehicles in complex situations.