Improvement of Computational Speed on the Embedded Implementation of the Localization Algorithm with Magnetic Sensor

초록

Magnetic sensor system is a representative localization system due to its simple structure and inexpensiveness. The magnetic sensor system is attached to the automatic driving vehicle such as PRT(personal rapid transit) and localize the vehicle by detecting the magnets which are embedded in the pre-defined path and computing the distance from the sensor to the magnet. The computational speed is an important concern because the localization algorithm should be implemented in real-time with relatively simple embedded microcontroller. Even though an algorithm is executed in the Windows system in short time, it does not guarantee that the algorithm is executed in the embedded system in time because there are some constraints using memories and floating point arithmetic units in the embedded system. In this study, efforts and methods to improve the computational speed of the localization algorithm are provided. The execution time measurements are also provided in order to represent the speed improvement of proposed methods.