Motion sickness is one of the most significant driving discomforts, significantly impacting safety, driving experience, and satisfaction. Drivers and passengers may experience different patterns of motion sickness due to differences in their tasks and...
Motion sickness is one of the most significant driving discomforts, significantly impacting safety, driving experience, and satisfaction. Drivers and passengers may experience different patterns of motion sickness due to differences in their tasks and the resulting movement information they perceive. Drivers can anticipate driving information by directly controlling the direction and speed of the vehicle, which aligns with the actual movement perceived through visual and vestibular systems, resulting in relatively less motion sickness. In contrast, passengers lack controllability, making it difficult for their brains to anticipate driving information. This leads to sensory conflicts between perceived or expected movements and actual vehicle movements, which can trigger motion sickness.
This issue is becoming increasingly important with the widespread adoption of electric vehicles (EVs) and autonomous vehicles (AVs), potentially acting as a barrier to the user experience in the adoption process of new vehicle technologies and services. The acceleration profiles of EVs differ from those of conventional internal combustion engine vehicles, causing discrepancies between expected and actual vehicle movements. Meanwhile, in AVs, increased driver freedom and non-driving-related activities reduce movement perception and prediction, potentially increasing the likelihood of motion sickness.
Against this background, it is crucial to develop systems that can monitor and predict passenger states in real-time to prevent and alleviate driving discomfort in future mobility environments. Considering the sustainability and feasibility of such systems, non-invasive technologies for measuring and predicting motion sickness-related states are ultimately required. To this end, designing human-vehicle interaction that is appropriate for passenger states is also important.
Previous studies on motion sickness in vehicles have evaluated motion sickness using data from subjective questionnaires with single or multiple items, signal-based physiological measurements, and vehicle movement (kinetic dynamics). However, these approaches have shown limited success in real-time prediction due to individual variability in symptom expression and the dynamic nature of driving environments. Additionally, subjective indicators present challenges in collecting real-time continuous data and interfere with non-driving-related activities, while also failing to account for substantial individual differences in symptom manifestation and severity perception. Meanwhile, quantitative indicators are either invasive or provide only average severity estimates. To overcome these limitations, it is essential to develop passenger motion sickness monitoring systems that are applicable to real-world autonomous driving environments. These systems should reflect the relationships between self-reported severity, serving as the ground truth, and non-invasively collected data through physiological sensors and vision-based methods.
Accordingly, this research aims to develop real-time monitoring and prediction models for passenger motion sickness and design mitigation strategies for the driving environments through the following three objectives. First, to establish an experimental framework to collect and analyze the data acquired from real vehicle driving environments, thereby identifying the occurrence of motion sickness by symptoms. Second, to develop real-time monitoring and prediction models by integrating subjective questionnaires, physiological measure, and vision-based facial changes. This enables the proposal of non-invasive systems that can detect the onset of motion sickness and worsening symptoms. Third, to design interfaces and mitigation strategies that adaptively respond to passenger motion sickness states and explore human-vehicle interaction in AV environments.
Chapter 3 introduces the experimental methodology and data collection process in a real-driving environment. Multimodal data, including subjective ratings, physiological signals, and vision-based observations, were gathered. Chapter 4 proposes a real-time monitoring system that utilizes the empirical data to predict passenger motion sickness status by symptoms. The model integrates self-report, physiological, and computer vision-based facial features to enable robust non-invasive prediction. Chapter 5 explores mitigation strategies and interface design solutions to reduce motion sickness in AVs. Context-aware feedback and adaptive interaction designs are proposed, along with evaluation metrics for their effectiveness in alleviating discomfort.
This study contributes theoretically and methodologically by identifying relationships between major motion sickness symptoms, physiological changes, and vision-based facial changes, and by developing motion sickness measurement methodologies encompassing mild to severe symptoms in real driving environments. Most significantly, by ergonomically understanding the relationship between the accuracy of self-report methods and the practicality of non-invasive measurement methods and proposing monitoring systems based on this understanding, the study established a foundation for continuously detecting passenger motion sickness states and implementing appropriate mitigation measures while passengers engage in various non-driving-related activities in future AVs. This is expected to substantially enhance the user experience and acceptance of AVs.