Lateral control with neural network head roll prediction model for motion sickness incidence minimisation in autonomous vehicle

Abstract

Generally, the passengers of an autonomous vehicle suffer substantial motion sickness (MS) compared to the driver. This particularly occurs during cornering as the passengers are inclined to tilt their heads in the direction of lateral acceleration whereas the driver tends to tilt their head in the opposite direction. Therefore, it is crucial for the passengers to reduce the head roll angle towards the direction of lateral acceleration to decrease the susceptibility to MS. This study proposed a lateral control approach based on the head roll angle which was estimated by head roll prediction models to reduce the severity of MS in an autonomous vehicle. The prediction models were developed via the Artificial Neural Network (ANN) technique. A Proportional-Integral (PI) controller was implemented to produce a corrective wheel angle based on the predicted head roll angle responses of the driver and passenger. The corrective angle caused a decrease in the lateral acceleration. The decrement in lateral acceleration then reduced the passenger’s head roll angle towards the direction of lateral acceleration. The findings indicated that the suggested control approach was capable to decrease the MS Incidence (MSI) index by 5.97% over a single lap and 14.48% over 10 laps.