Accurate simulation and validation of advanced driver assistance systems requires accurate sensor models. Modeling automotive radar is complicated by effects such as multipath reflections, interference, reflective surfaces, discrete cells, and attenuation. Detailed radar simulations based on physical principles exist but are computationally intractable for realistic automotive scenes. This paper describes a methodology for the construction of stochastic automotive radar models based on deep learning with adversarial loss connected to real-world data. The resulting model exhibits fundamental radar effects while remaining real-time capable.

The virtual validation of automated driving functions requires meaningful simulation models of environment perception sensors such as radar, lidar, and cameras. There does not yet exist an unrivaled standard for perception sensor models, and radar especially lacks modeling approaches that consistently produce realistic results. In this paper, we present measurements that exemplify challenges in the development of meaningful radar sensor models. We highlight three major challenges: multi-path propagation, separability, and sensitivity of radar cross section to the aspect angle. We also review previous work addressing these challenges and suggest further research directions towards meaningful automotive radar simulation models.

Simulation-based testing is seen as a major requirement for the safety validation of highly automated driving. One crucial part of such test architectures are models of environment perception sensors such as camera, lidar and radar sensors. Currently, an objective evaluation and the comparison of different modeling approaches for automotive lidar sensors are still a challenge. In this work, a real lidar sensor system used for object recognition is first functionally decomposed. The resulting sequence of processing blocks and interfaces is then mapped onto simulation methods. Subsequently, metrics applied to the aforementioned interfaces are derived, enabling a quantitative comparison between simulated and real sensor data at different steps of the processing pipeline. Benchmarks for several existing sensor models at a concrete selected interface are performed using those metrics by comparing them to measurements gained from the real sensor. Finally, we outline how metrics on low-level interfaces can correlate with results on more abstract ones. A major achievement of this work lies within the commonly accepted interfaces and a common understanding of real and virtual lidar sensor systems and, even more important, an initial guideline for the quantitative comparison of sensor models with the ambition to support future validation of virtual sensor models.

Validating safety is an unsolved challenge before autonomous driving on public roads is possible. Since only the use of simulation-based test procedures can lead to an economically viable solution for safety validation, computationally efficient simulation models with validated fidelity are demanded. A central part of the overall simulation tool chain is the simulation of the perception components. In this work, a sequential modular approach for simulation of active perception sensor systems is presented on the example of lidar. It enables the required level of fidelity of synthetic object list data for safety validation using beforehand simulated point clouds. The elaborated framework around the sequential modules provides standardized interfaces packaging for co-simulation such as Open Simulation Interface (OSI) and Functional Mockup Interface (FMI), while providing a new level of modularity, testability, interchangeability, and distributability. The fidelity of the sequential approach is demonstrated on an everyday scenario at an intersection that is performed in reality at first and reproduced in simulation afterwards. The synthetic point cloud is generated by a sensor model with high fidelity and processed by a tracking model afterwards, which, therefore, outputs bounding boxes and trajectories that are close to reality.

Safety validation tests of automated driving (AD) use simulated environments and perception sensor models. To achieve the level of fidelity needed for safety approval, such sensor simulations can be physically based. For formulating requirements for sensor models to be used in such test frameworks, the extent to which they must include physical effects should be determined. One approach is to clarify their relevance for following processing steps like object detection or mapping. But at first, an analysis is needed to determine, which effects are relevant and if they can be implemented at all. In this work, we focus on one lidar sensor and analyze its observable real world sensor behavior to derive the possible effects, physically based lidar sensor models can include. Consequently, we describe environmental parameters that could be considered to influence physically based lidar sensor models. By investigating the specifications given by the manufacturer with own measurements, we show that some of them should be implemented in a dynamic manner. In conclusion, we enable to formulate detailed requirements for sensor models, as their actual possible fidelity is presented.