Safe navigation of autonomous vehicles relies on high integrity localization system based on a state estimation method, which strongly depends on the choice of parameters, such as the measurement noise covariance matrix. To avoid filter divergence, a fault detection and exclusion (FDE) procedure is also required. Selecting appropriate thresholds for the FDE step is challenging and affects the accuracy and the integrity of localization. This paper presents a multi-sensor data fusion based on the Information Filter (IF) with an auto-tuning method that relies on Bayesian Optimization (BO). The proposed method aims to optimize the measurement covariance matrix and the FDE thresholds. BO is well suited for non-convex and stochastic cost functions with local minima. A novel objective function is also designed to improve the accuracy and guarantee the integrity of the estimates, by ensuring consistent uncertainty regions. The objective function consists of two terms: one related to the error and the other to the quantile. The proposed approach is evaluated with experimental data from a vehicle equipped with wheel speed sensors, fused with Global Navigation Satellite System (GNSS) pseudorange measurements.

The navigation of intelligent vehicles relies on high integrity localization system capable to bound the estimation errors. This paper introduces a zonotopic and Gaussian Kalman filter in informational form for multi-sensor data fusion and confidence domain computation. By integrating stochastic and set membership uncertainties, the proposed filter ensures accurate localization with a non pessimistic confidence domain, thus addressing the challenges posed by traditional techniques. Taking advantage of the informational form, a fault detection and exclusion step is added to enhance filter robustness. Following a zonotope reduction step, a confidence domain computation, considering both Gaussian and zonotopic uncertainties, is proposed in the context of intelligent vehicles. The accuracy and integrity of the approach are assessed using experimental data, including the fusion of GPS and Galileo pseudoranges with camera measurements after a map matching step. Additionally, a comparative analysis is conducted with the classical Kalman filter.