Enhancing AEB Systems with Digital Twin Evaluation in Urban Environments

Autonomous Emergency Braking (AEB) systems are pivotal in preventing collisions. However, their effectiveness significantly diminishes in urban environments, where complex scenarios and adverse weather conditions complicate the estimation of distances to other road users and adaptation to variable road-tire friction coefficients. Therefore, it is critical to conduct a comprehensive analysis of the inaccuracies in distance and road condition estimations to enhance AEB performance. Traditional AEB testing methods, which involve reproducing critical traffic situations and altering road conditions, pose significant risks and challenges. To address these issues, our study introduces a digital twin-based evaluation methodology that replicates our test vehicle, complete with realistic driving dynamics and physical sensor models. Our testing protocols conform to the European New Car Assessment Programme (EU NCAP) standards for AEB Car-to-Pedestrian scenarios, with extensions to accommodate complex traffic and weather situations. Moreover, we adjust different roadway conditions from dry to icy in simulation. Our results determine the error tolerance and highlight the necessary precision for perception systems. Furthermore, our research illuminates the primary challenges associated with employing a digital twin in the development of autonomous driving functionalities and suggests possible implications.