This paper presents a method for fusing two maps of an environment: one estimated with an application of the Simultaneous Localization and Mapping (SLAM) concept and the other one known a priori by a vehicle. The goal of such an application is double: first, to estimate the vehicle pose in this known map and, second, to constrain the map estimate with the known map using an implementation of the local maps fusion approach and a heterogeneous mapping of the environment. This article shows how a priori knowledge available in the form of a map can be fused within an EKF-SLAM framework to obtain more accuracy on the vehicle poses and map estimates. Simulation and experimental results are given to show these improvements.