A Survey on Hardware Models for Virtual Development of Sensor Systems for Autonomous Driving

The research and development of autonomous driving systems has seen a significant increase in recent years. However, for high levels of automation, specifically SAE Levels 3, 4, and 5, the vehicle requires a perfect understanding of the surrounding environment. This is accomplished by a perception system consisting of sensors, such as camera, radar and lidar. The reliability of such systems is crucial in ensuring the safety requirements of autonomous vehicles. To develop, train, and validate the functionality of these systems, virtual models of the environment and hardware components are essential. This paper provides an extensive survey on various hardware models used for virtual development of perception sensor systems in autonomous driving. In this survey, we have examined both open source and proprietary approaches documented in the literature. Our analysis covers a range of aspects, including available environment models, sensor models, controller models and standardized model interfaces. By comparing different approaches, we have evaluated their availability, essential features, and their usage in the scientific community. Based on the results, we have identified three critical development aspects that must be addressed in the near future to improve and extend the development of perception sensor systems for autonomous driving.