Research on Application of Machine Learning in Pedestrian Headform to Bonnet Top Tests

Abstract

AbstractPedestrian vehicle safety accreditation is a mandatory requirement for all cars marketed in the European Union. Therefore, when researching and developing a new car model, all automakers must study to design pedestrian protection in collisions. In the process of research and development as well as pedestrian safety accreditation, there are a lot of impacting tests to evaluate pedestrian safety. As for bonnet top area, there are at least 18 tests of headform impact to bonnet top to evaluate pedestrian head safety. Since the number of positions selected for testing is finite and discrete, it is difficult to accurately conclude the pedestrian safety level for every position on the bonnet surface area. In this study, the four models of machine learning (ML) algorithm Linear regression (LIN), Support Vector Machine Regression (SVR), Extreme Gradient Boosting (XGB) and Artificial Neural Network (ANN) are applied to predict the safety level of every position on the bonnet top surface based on the data obtained from testing at several positions. In order to assess the predictive performance of ML model, the error metrics are used such as root mean squared error (RMSE), mean absolute error (MAE) and coefficient of determination (R2). The GridSearchCV algorithm is also used to find the best hyperparameter of ML model. The results of this study will be very useful for the development of solutions to vehicle safety tests for pedestrians. LIN is the best model in predicting the HIC value of headform and bonnet collision.KeywordsPedestrian head safetyMachine learningPrediction HIC valueHeadform impactor to bonnet top testsPedestrian vehicle safety accreditation