Trajectory Planning for Motion Sickness Mitigation in Autonomous Driving: Effect of Frequency Weighting and Road Three-Dimensionality

Autonomous driving is expected to significantly affect future mobility. While the technological aspects are often the focus, various subjective challenges such as perceived comfort also need to be addressed before autonomous vehicles are accepted by users. One of the key restrictions on the design is motion sickness, since autonomous vehicles are expected to increase its incidence and severity due to the shift of the user from driver to passenger. To maximise user acceptance a human-centred approach is necessary. In this work, the optimisation of trajectory planning as a trade-off between motion sickness and manoeuvre time is carried out by considering the Motion Sickness Dose Value (MSDV, defined in ISO 2631) within the cost function of an optimal control problem (OCP). All motion directions are considered for the computation of the MSDV, i.e. vertical, longitudinal and lateral, with different frequency weightings in each direction. The filters associated with the frequency weightings increase the dimension, i.e. number of states, of the OCP. Different approximations of the exact (high order) frequency weightings are employed and the analysis is carried out on three-dimensional tracks. The related OCP is solved using a direct approach in combination with a nonlinear programming solver. The results show that lower-order approximations of the weighting filters are sufficient for the computation of the trajectory and speed profiles that mitigate the motion sickness, while optimisation based on minimising acceleration RMS or jerk RMS leads to higher MSDV. The effect of three-dimensionality (slope and banking) is negligible in terms of calculated MSDV, for the selected tracks and vehicle model.