Graph-based SLAM based magnet map generation for magnetic guidance

Magnetic guidance is a reliable navigation method for intelligence vehicles and AGVs. To get the right and accurate location of vehicle for control, an accurate magnet map is needed. This paper proposed a new approach to build a magnet map by graph-based SLAM. Vehicle´s poses in the driving trajectory can be estimated by SLAM, for the magnetic sensor installed on the vehicle is detecting the relative position of vehicle and the magnetic markers embedded under the road during the vehicle driving, the markers´ positions can be calculated by the vehicle poses estimated by SLAM. But the result of normal SLAM may have loop closure problem in large and complex environment due to measurement error, graph-based SLAM will be the solution to this problem, and conversely, the relative position of magnetic markers measured by the sensor can be used as loop closure constraints for pose optimization. Experiments in large-scale outdoor scenario have been conducted and the magnet map result shows its feasibility.