Analysing the Effects of Large Rearward Driver Seat Adjustments Using Real-World Crash Data and Stochastic Simulations with Multiple Human Body Models

Abstract Improvements in road and vehicle safety have led to fewer motor vehicle occupant fatalities and injuries over time. However, in many countries worldwide, this trend has stagnated. Among other measures, the introduction of automated vehicles (AVs) is expected to increase road traffic safety by reducing the number of crashes and reduce the severity of unavoidable collisions.Aside from the anticipated safety benefits, AVs enable novel seat configurations, since the person in the driver’s seat does not need to reach the controls when the car is driving itself. However, AVs must ensure occupant safety equivalent to conventional vehicles in unavoidable crashes. In a worst-case scenario, a misjudgement or system failure could lead to the AV being involved in a crash with the same severity as a conventional vehicle, but with the occupants in novel positions. The present thesis aims to assess this kind of fallback performance of the restraint systems for large rearward driver seat adjustments. Frontal collisions from an in-depth crash database were characterised and parametric distributions were fitted to key collision variables. In addition, other relevant distributions were obtained from literature. These were used in stochastic simulations with detailed human body models (HBMs) representing the drivers. Four HBMs (small female and average male THUMS, average female and average male VIVA+) were used, enabling comparisons under identical loading conditions. Initially, simulations with a conventional driving position were compared to the driver injuries obtained from the sample used to characterise the collisions for the simulations, achieving a replication accuracy of ±10% for all body regions except the head. Repeating the simulations with the driver’s seat moved rearwards revealed increased injury risks for all body regions except the head. However, the reduction in head injury risk was attributed to the airbag setup being overly stiff in the conventional driving position. Notably, all models predicted considerable risk increases for the lower extremities, with the small female human body model exhibiting the highest injury risk.In summary, while novel seat configurations pose new challenges, the presented approach can be seen as one element to mitigate potential adverse effects on occupant safety, before they emerge in real-world crashes. By integrating real-world data with stochastic simulations, this study provides crucial insights into injury mechanisms and sets the foundation for more accurate and inclusive safety evaluations, which are essential for the safe design of AVs and the reduction of injury risks. Future research should focus on improving crash data collection, refining injury risk functions, and aligning injury criteria across human body models to further enhance predictive safety assessments.