Intelligent driving multi-source information fusion fault-tolerant system based on an improved Sage-Husa adaptive extended Kalman filter

Abstract

With the rapid advancement of intelligent driving technology, multi-source information fusion has emerged as a critical focus in environmental perception research. This technology not only reliably captures environmental data across diverse conditions but also enables cross-sensor data complementarity. To address the fusion deviation resulting from changes in sensor performance due to environmental variations, this paper proposes a multi-source information fusion system utilizing the Sage-Husa adaptive extended Kalman filtering (SHAEKF) algorithm. First, we construct a multi-source information fusion system grounded in vehicle motion models. Then, a fading factor is incorporated into the SHAEKF algorithm, such that the estimation error is effectively reduced and the risk of filter divergence is mitigated. Finally, a dynamic threshold fault-tolerance module is proposed based on Euclidean distance and motion equations to identify and eliminate erroneous sensor information. Based on a series of simulation experiments, the improved SHAEKF algorithm demonstrates higher estimation accuracy and robustness than the original SHAEKF algorithm.