Prediction of occupant’s head movement during slalom driving via artificial neural network with multiple training algorithms

Abstract

Autonomous vehicles are one of the future transportation technologies across the globe. However, autonomous vehicles have some setbacks and one of the setbacks is motion sickness. Occupant’s comfort level plays a vital role in the development of an autonomous vehicle. The motion sickness occurs due to the head movement of the driver tends to tilt against lateral acceleration but towards centripetal force; the head movement of the passenger tends to tilt against centripetal force but towards lateral acceleration. In addition, the method to develop and increase the comfort level is to monitor the head movement of the occupants during slalom driving. Nevertheless, it is inappropriate to attach sensors on occupants while traveling due to discomfort and dissatisfaction of driving experience. Hence, this study proposes prediction model of occupant’s head movement via Artificial Neural Networks. The data is taken from previous work from research. The experiment is carried out to collect the response data of lateral acceleration and the head movements of the occupants. This research also presents the model developed in MATLAB by implementing experimental data as parameter into two different training algorithms, Levenberg-Marquardt algorithm and Bayesian Regularization algorithm.