Simultaneous landmarks detection and data association in noisy environment for map aided localization

In this paper, we make an analysis of map aided localization systems (MAL) in noisy outdoor environments. In a first time, we present the main sensors used in those approaches and their behavior in noisy environments. We notice that sensors don´t offer the same performance. In MAL approaches, those sensors are used to detect landmarks. This detection process constitutes the first step to locate the vehicle. The second step is a data association phase. In the classical systems, those two steps are separated and we remark they have some difficulties to work in noisy environments. We propose here to process them simultaneously using a focusing method. This last one allows the system to return only the landmarks compatible with what we are looking for. Thus, only the desired landmark is returned by the system. This is achieved by taking into account all features of the searched landmark in the detection process. Some experimentations of automatic vehicle guidance were made in outdoor environment under a strong downpour of hail showing the benefit of the proposed approach.