Fahrzeugsicherheit

Minimizing occupant loads in vehicle crashes through reinforcement learning-based restraint system design: assessing performance and transferability

“…Subsequently, structural responses are often highly nonlinear and sensitive (Schuhmacher, 2020). During the development process, simultaneous optimization of the vehicle structure and the restraint system is performed with the aim of ensuring the integrity of the structure during a crash and reducing the loads on anthropomorphic test devices (ATD) (Schuhmacher, 2020;Gonter et al, 2021). For optimizing mechanical system behavior global optimization algorithms (e.g., genetic algorithms), local optimization algorithms (e.g., gradient methods), but also heuristic-based approaches can be applied (Schuhmacher, 2020).…”

“…The ATD loads to be minimized by adjusting the restraint system configuration are typically part in either statutory regulations such as FMVSS 208 or consumer protection guidelines such as the Euro-NCAP (Gonter et al, 2021). Relevant values are, for instance, the head injury criterion in an interval of 15 ms (HIC15), neck injury criterion (Nij) or the maximal chest acceleration.…”

“…With more than one million people killed on the roads worldwide in each year over the past decades, the consideration of safety-relevant aspects to better protect passengers and pedestrians will continue to play a crucial role in the development process of new vehicle generations (Gonter et al, 2021). To improve the crashworthiness of vehicles, structures such as side sills, longitudinal and cross members, but also the restraint system, which is particularly relevant for occupant protection, are optimized regarding the best possible mitigation of injury risks.…”

“…To improve the crashworthiness of vehicles, structures such as side sills, longitudinal and cross members, but also the restraint system, which is particularly relevant for occupant protection, are optimized regarding the best possible mitigation of injury risks. The considerations to address the optimization of mechanical system behavior are carried out virtually with the aid of numerical finite element (FE) simulations since physical tests of individual components or entire prototypes are highly time-and resource-intensive (Schuhmacher, 2020;Gonter et al, 2021). This enables concepts to be evaluated more quickly which results in more agile and efficient development processes.…”

“…For passive vehicle safety, this necessitates conducting numerous investigations to disentangle complex relationships and dependencies. The resulting large number of required simulations, coupled with the intricacy and size of FE models, also render virtual assessments computationally-and thus cost-intensive (Schuhmacher, 2020;Gonter et al, 2021). To further digitalize and increase the efficiency of modern development processes, recent focus has shifted towards datadriven approaches and concepts (Langen et al, 2022).…”