Lateral dynamics modeling: A black-box modeling of the dynamics of a car

Abstract

[ES] En este documento se describe: -Un estudio bibliográfico acerca de los distintos modelos de dinámica lateral que pueden ser implementados en un vehículo autónomo. -Los tests que deben ser realizados para la identificación de un modelo de caja negra, lo cual incluye el vehículo utilizado, las instalaciones, el diseño de los tests y los datos obtenidos. -El proceso de identificación de los parámetros relevantes del modelo (en orden cronológico). -La presentación del modelo final y su comparación con otros modelos desarrollados previamente.

[EN] To control a vehicle, in this case a car, it is necessary to know in advance how the car is going to behave. The trajectory of the car can be determined if the amount of degrees that the steering wheel turns is known. This is called the lateral dynamics of the car. The lateral dynamics should be synthesized in a model, this is, a set of equations. Once the model has been develop, the state of the car (relevant variables) will be known without any need of further testing. These models include a set of parameters, that can have a physical meaning or not, whether the model is a Gray Box Model or a Black Box Model, respectively. For the purpose of obtain- ing these parameters some tests should be performed on the car, which must be equipped with different sensors needed in order to obtain or calculate these relevant parameters. Once the test are completed, it is necessary to perform a precise analysis of the acquired data to obtain the coefficients that ultimately define the system. After the model is defined and validated, it is possible to compare it with other models to test its accuracy. In the following pages, the reader will find all the different issues that have been considered for developing a self-driving car lateral dynamics model.