This paper presents a multi-sensor fusion strategy able to detect the spurious sensors data that must be eliminated from the fusion procedure. The used estimator is the informational form of the Kalman Filter (KF) namely Information Filter (IF). In order to detect the erroneous sensors measurements, the Kullback-Leibler Divergence (KLD) between the a priori and a posteriori distributions of the IF is computed. It is generated from two tests: One acts on the means and the other deals with the covariance matrices. Optimal thresholding method based on a Kullback-Leibler Criterion (KLC) is developed and discussed in order to replace classical approaches that fix heuristically the false alarm probability. Multi-robot systems became one of the major fields of study in the indoor environment where the environmental monitoring and the response to crisis must be ensured. Consequently, the robots required to know precisely their positions and orientations in order to successfully perform their mission. Fault detection and exclusion (FDE) play a crucial role in enhancing the integrity of localization of the multi-robot team. The main contributions of this paper are: - developing a new method of sensors data fusion that tackle the erroneous data issues, - developing a Kullback-Leibler based criterion for the threshold optimization, - Validation with real experimental data from a group of robots.