An Adaptive Fault-Tolerant Multisensor Navigation Strategy for Automated Vehicles

Abstract

This paper presents a novel adaptive fault-tolerant multisensor navigation strategy for automated vehicles on the automated highway system. To ensure safe and reliable running, the main concern is how we can realize sensor fault detection, isolation, and accommodation. The proposed strategy adopts a hybrid decentralized filter architecture that fuses multiple sensors, including the strapdown inertial navigation system, the carrier phase-differential Global Positioning System (GPS), the wheel encoder, and the electronic compass. Through the integration of multiple redundancy sensors, the navigation system has fault-tolerant potential. The integrated filter model is first established in detail. Then, a highly fault-tolerant navigation filter iterative algorithm is proposed. The proposed algorithm realizes fault tolerance by utilizing fuzzy logic to update the local filter based on a relative degree of mismatch by simultaneously considering sensor failure degree and to make fault isolation decisions. To compare and evaluate the fault-tolerant performance, several traditional navigation methods are also investigated. Simulation results demonstrate that the proposed integrated navigation method can effectively detect, isolate, and accommodate a variety of hard and soft sensor failures, which may even simultaneously occur.