Vehicle localization based on odometry assisted magnetic matching

Magnetic is a promising technical for positioning in the GNSS denied scenarios such as parking lot and tunnel, since the magnetic field exists everywhere on the earth. In this paper, for vehicle navigation, an odometry assisted magnetic matching algorithm is proposed, in which the magnetometer and the odometry of vehicle are used. The proposed scheme consists of two phases: offline phase and online phase. During the offline phase, the magnetic map is constructed which includes the magnetic field intensities and the coordinates of Reference Points, which are distributed uniformly along the road. During the online phase, the position is estimated by matching the magnetic field intensities in a certain distance window to the magnetic map constructed in the offline phase. To overcome the mismatch and error fluctuation problem, odometry is also used to evaluate the result. Experiments based on a mini bus showed that the performance of proposed scheme is accurate (2a:4.8m) and reliable (max error: 13m).