Simulation Framework for Testing Train Navigation Algorithms Based on 9-DOF-IMU and Tachometers

The integration of candidate sensors such as a microelectromechanical system-based inertial measurement unit (MEMS IMU) for localization and navigation has been widely explored for vehicular, pedestrian, and robotics applications. They also arise as candidate sensors in the train navigation context, and their integration in safety-relevant applications has become a prominent area of interest. This requires sensor modeling in a specific context, the search of new information sources (e.g., digital maps), the development of navigation algorithms, and the testing of the system. In this context, this article proposes a one-stop simulation framework for testing train navigation algorithms that uses nine degrees of freedom (DOFs) IMU and tachometers. The simulation framework can generate synthetic signals based on the dynamic model of a train, digital maps, speed profiles, and a sensor error model. It also uses real signals recorded from experimental measuring campaigns. Train navigation algorithms are then executed. Finally, the performance of the obtained results is evaluated. The main contribution made by this framework is its flexibility when used to evaluate different sensors, train configurations, tracks, mechanization methods, navigation algorithms, and so on during different stages of the development process. Simulation framework use cases are presented in this article, where several mechanization methods and train navigation algorithms are tested. Finally, the experimental data are recorded and included in the presented simulation framework to evaluate an example of a train navigation algorithm based on a 9-DOF-IMU and tachometers. The mean differences in the synthetic and experimental IMU signals are around 3.52% in accelerometers, below 0.01% in gyroscopes, and 10.16% in magnetometers in the acceleration sections (i.e., when the train is applying torque).