Most of the existing mobile robot localization solutions are either heavily dependent on pre-installed infrastructures or having difficulty working in highly repetitive environments which do not have sufficient unique features. To address this problem, we propose a magnetic-assisted initialization approach that enhances the performance of infrastructure-free mobile robot localization in repetitive featureless environments. The proposed system adopts a coarse-to-fine structure, which mainly consists of two parts: magnetic field-based matching and laser scan matching. Firstly, the interpolated magnetic field map is built and the initial pose of the mobile robot is partly determined by the k-Nearest Neighbors (k-NN) algorithm. Next, with the fusion of prior initial pose information, the robot is localized by laser scan matching more accurately and efficiently. In our experiment, the mobile robot was successfully localized in a featureless rectangular corridor with a success rate of 88% and an average correct localization time of 6.6 seconds.