Desenvolvimento de modelo de rede neural artificial para previsão do desempenho subjetivo da sensação de direção a partir de métricas objetivas de dinâmica veicular lateral

Abstract

This study presents the development of an artificial neural network (ANN) model to predict the subjective performance of vehicles based on objective metrics of lateral vehicle dynamics and steering feel. Different vehicle segments were characterized through simulations and clustered by similarity, with a representative sample subsequently selected for subjective evaluation. These evaluations were carried out in a driving simulator, in which ratings were assigned to parameters related to directional response, roll dynamics, and steering feel, according to a predefined comparative scale. The performance of the ANN was compared with that of other classical machine learning models, proving to be superior. In addition, a metric importance analysis was conducted, identifying the factors with the greatest influence on the model’s predictive capability. The results confirm the potential of the approach to support vehicle development by aligning objective metrics with driver perception, although improvements are still needed in the prediction of specific parameters.