Multisensor Fusion Localization using Extended H∞Filter using Pre-filtered Sensors Measurements

Abstract

Localization system is considered one of the main components of an intelligent vehicle, and it presents a criticality from the safety and performance point of view, which requires high accuracy and precision. However, the performance of these systems is affected by sensor limitations and, more importantly, noise. The localization problem can be solved by fusing the information from multiple sensors, and the multisensor fusion problem is typically solved using probabilistic methods. Among these methods, Kalman filters are the most widely used. Although Kalman filters are the most widely used, the noise models of the sensors fused are assumed to be Gaussian, and the noise statistics must be known beforehand, leading to a sub-optimal performance in systems with unknown noise characteristics. Accordingly, this paper presents a robust localization system, based on the extended $\\mathcal{H}_{\\infty}$filter, in order to solve the localization problem in intelligent self-driving vehicles by fusing different localization sources. The proposed approach is validated through several real-world experiments under different scenarios, and the performance of the proposed method is compared with the Extended Kalman filter under the same conditions. The obtained results demonstrate that the proposed approach outperforms the Extended Kalman filter, and validates the use of the extended $\\mathcal{H}_{\\infty}$filter as a multisensor fusion approach for localization systems.