Injury Assessment in Non-Standard Seating Configurations in Highly Automated Vehicles Using Digital Twin and Active Learning

Human-driven vehicles are going to be replaced by highly automated vehicles as one of the future mobility trends. Even though highly automated vehicles’ active safety systems can protect against vehicle-to-vehicle accidents, the traffic mix between human-driven vehicles and highly automated vehicles is still a potential source of vehicle collisions. Additionally, occupants in highly automated vehicles will be passengers not necessarily dealing with driving anymore, so there will be a considerable number of non-standard seating configurations. Those configurations are not able to be assessed for safety by hardware testing due to their number, variability and complexity. The objective of the paper is the development of a fast virtual approach to identify the passengers’ injury risk in non-standard seating configurations under multi-directional impact scenarios and severity. We deploy the concept of surrogate modeling, where we introduce a digital twin for the expected automated vehicle interiors. Non-standard seating configurations are represented by a simplified model of four seats located in the vehicle. These seats are occupied by a previously developed scalable human body model representing passengers of variable anthropometry. Thanks to the vehicle interior simplification and the hybrid human body model, thousands of simulations describing the impacts identified can be run. Based on the numerical simulations describing impact scenarios, a fast and lean artificial intelligence (AI) model actively learns a digital twin to approximate injury risk predictions for a huge number of possible crash scenarios fast. An impact scenario concerns a seating configuration occupied by up to 4 passengers (a seat can be empty) of variable height, weight, age and gender and crash direction and severity (velocity). Behind AI, the machine learning method uses supervised classifiers that are trained to predict injury severity based on the given input. There are 8 trained classifiers per passenger, each one for a body segment, where multi (predicting 4 injury severity levels) and binary (predicting injury or non-injury levels) classifications are compared. The machine learning accuracy is compared by the Mattheus correlation coefficient, where the presented digital twin AI approach reasonably approximates the numerical solution.