By 2050 a large proportion of the cars on our roads will be self-driving and completely automated. We will no longer be driving these vehicles, but will be transported comfortably as passengers. We will be able to indulge in all sorts of media items in our vehicles, do work, or even just relax and sleep. Indeed, these fully automated vehicles will have a clear positive impact on everybody’s lives. That is, if people do not become too motion sick to enjoy the ride. It is known that drivers of vehicles do not get motion sick because they are in control of the vehicle and, hence, can anticipate upcoming motions. However, many passengers, which we will all eventually become, do experience motion sickness. This is particularly an issue when their eyes are off the road and when they are engaged in other activities. With a rising prospect of motion sickness, these activities may no longer seem attractive. Moreover, motion sickness increases workload and decreases cognitive performance, which means that those wishing to use their commuting time in cognitively demanding activities will be less productive in them. With full automation, it is hoped that vehicle control systems can be optimised to reduce sickening motions to the lowest feasible level, whilst also achieving adequate vehicle performance in terms of, for instance, journey time. However, at this moment we don’t yet have a good model of motion sickness that would enable such optimization. For example, route planning algorithms generally optimise for the shortest time, while some have recently started to optimise for the least polluting route. For this optimisation, the algorithm needs to have, amongst other things, traffic information, the roads, their lengths and legally allowable speeds, with all such information encapsulated in a general mathematical model. There is, however, no analogous model of motion sickness that can be used to optimise our vehicle’s behaviour for the lowest passenger sickness incidence. A mathematical model in this sense is a set of equations that tell us how sickening a certain pattern of motion will be. A pattern of motion could be the vehicle taking a turn, switching a lane, stopping at a traffic light; anything that makes the vehicle change speed or direction. By being able to predict how much sickness will result from certain manoeuvres, the vehicle can be programmed to perform these manoeuvres in the least sickening manner. One problem with developing mathematical models for motion sickness minimisation is that there is a great variability in how motion sickness manifests itself in individuals. This means that the motion sickness symptoms, their accumulation and even the nature of the motions that cause sickness are highly individual. Therefore, any algorithm meant for vehicle control must take the individual as the subject of its concern. Prior to this thesis, the individual was not seen as a feasible unit of study. Instead, literature mainly focused on group-level responses. However, because motion sickness is so variable, it is likely that optimising group-averaged criteria, will not optimise for group-averaged comfort. Instead, individualisation is needed. This need directly shaped the objective of this thesis, which was to understand and model motion sickness accumulation and its individual differences for the comfortable control of automated vehicles...