Radial basis function neural network for head roll prediction modelling in a motion sickness study

Abstract

Motion Sickness (MS) is the result of uneasy feelings that occurs when travelling. In MS mitigation studies, it is necessary to investigate and measure the occupant’s Motion Sickness Incidence (MSI) for analysis purposes. One way to mathematically calculate the MSI is by using a 6-DOF Subjective Vertical Conflict (SVC) model. This model utilises the information of the vehicle lateral acceleration and the occupant’s head roll angle to determine the MSI. The data of the lateral acceleration can be obtained by using a sensor. However, it is impractical to use a sensor to acquire the occupant’s head roll response. Therefore, this study presents the occupant’s head roll prediction model by using the Radial Basis Function Neural Network (RBFNN) method to estimate the actual head roll responses. The prediction model is modelled based on the correlation between lateral acceleration and head roll angle during curve driving. Experiments have been conducted to collect real naturalistic data for modelling purposes. The results show that the predicted responses from the model are similar with the real responses from the experiment. In future, it is expected that the prediction model will be useful in measuring the occupant’s MSI level by providing the estimated head roll responses.