In the future, mobility will be strongly shaped by the increasing use of digitalization. Not only will individual road users be highly interconnected, but also the road and associated infrastructure. At that point, a Digital Twin becomes particularly appealing because, unlike a basic simulation, it offers a continuous, bilateral connection linking the real and virtual environments. This paper describes the digital reconstruction used to develop the Digital Twin of the Test Area Autonomous Driving-Baden-Württemberg (TAF-BW), Germany. The TAF-BW offers a variety of different road sections, from high-traffic urban intersections and tunnels to multilane motorways. The test area is equipped with a comprehensive Vehicle-to-Everything (V2X) communication infrastructure and multiple intelligent intersections equipped with camera sensors to facilitate real-time traffic flow monitoring. The generation of authentic data as input for the Digital Twin was achieved by extracting object lists at the intersections. This process was facilitated by the combined utilization of camera images from the intelligent infrastructure and LiDAR sensors mounted on a test vehicle. Using a unified interface, recordings from real-world detections of traffic participants can be resimulated. Additionally, the simulation framework's design and the reconstruction process is discussed. The resulting framework is made publicly available for download and utilization at: https://digit4taf-bw.fzi.de The demonstration uses two case studies to illustrate the application of the digital twin and its interfaces: the analysis of traffic signal systems to optimize traffic flow and the simulation of security-related scenarios in the communications sector.

Perceptual fidelity plays a critical role in synthetic imagery for closing the sim-to-real gap in computer vision. The hypothesis that perceptual similarity between artificial and real images can be quantified using learned metrics such as LPIPS predicts downstream ML performance. In this paper, we support this hypothesis by presenting a method for recreating high-fidelity traffic scenes grounded in real-world data and comparing these scenes with real ones. We reconstruct scenes from the KITTI dataset using open geographic data sources, including OpenStreetMap and federal geospatial repositories. The environment layout, static infrastructure, and camera trajectories are derived from real-world constraints, while time-of-day and lighting conditions are matched through solar position calculations based on KITTI timestamps and GPS data. These scenes are rendered in Unreal Engine 5 using physically based materials and real-time global illumination (Lumen). Our approach achieves LPIPS scores up to two times lower compared to Virtual KITTI v2, indicating stronger perceptual alignment with the original dataset. In addition, we evaluate semantic segmentation performance, observing that models performing on our synthetic scenes exhibit smaller domain-induced degradation. These results substantiate the use of perceptual metrics as proxies for model transferability and underscore the importance of realism-focused simulation in bridging the sim-to-real domain gap.

Accurately representing road surfaces is crucial for automotive simulation and virtual sensor data generation because it directly impacts vehicle dynamics, sensor performance, and overall simulation realism. The precision of road surface models is critical for ensuring the accuracy and reliability of simulated driving scenarios, affecting various aspects such as tire-road interactions, vehicle handling, and the behavior of Advanced Driver Assistance Systems (ADAS) and autonomous vehicles. This paper presents a comprehensive material model that incorporates the complex properties of asphalt roads, including composition, binder content, aggregate gradation, age, and traffic load. By adjusting these parameters, our model dynamically generates shader materials that accurately replicate real-world asphalt surfaces' texture, color, and reflectance characteristics, which is essential for precise virtual sensor simulations. Moreover, our material model goes beyond visual fidelity to influence vehicle dynamics in traffic simulations. By introducing an additional physical material layer affected by environmental conditions, our model accurately simulates the interaction between vehicles and road surfaces, enhancing the realism and accuracy of simulated driving scenarios. The model also integrates smoothly with OpenDRIVE road definitions, allowing modifications of surface geometry to reflect road damage, repairs, and irregularities.

Collective perception has received considerable attention as a promising approach to overcome occlusions and limited sensing ranges of vehicle-local perception in autonomous driving. In order to develop and test novel collective perception technologies, appropriate datasets are required. These datasets must include not only different environmental conditions, as they strongly influence the perception capabilities, but also a wide range of scenarios with different road users as well as realistic sensor models. Therefore, we propose the Synthetic COllective PErception (SCOPE) dataset. SCOPE is the first synthetic multi-modal dataset that incorporates realistic camera and LiDAR models as well as parameterized and physically accurate weather simulations for both sensor types. The dataset contains 17,600 frames from over 40 diverse scenarios with up to 24 collaborative agents, infrastructure sensors, and passive traffic, including cyclists and pedestrians. In addition, recordings from two novel digital-twin maps from Karlsruhe and Tubingen are included. The dataset is available at https://ekut-es.github.io/scope

The popularity of environment simulations for fast assessment of autonomous driving functions is growing faster than ever before. But the creation of virtual environments to simulate real driving scenarios remains a challenge. Modeling objects by hand is a rather slow and expensive process. An alternative approach is procedural generation. Unfortunately, most methods produce only fictional environments. In this paper, we present a novel approach to procedurally generate environments for automotive simulations, that are based on real environment recordings and adress physical sensor specific reflection mechanisms. We use OpenDRIVE to generate the layout and road network and augment it with open source data from OpenStreetMap for objects surrounding the road. In addition, we present a workflow for rapid reconstruction of building facades, which allows us to create realistic and immersive environments for simulations. Our approach is based on the Unreal Engine 4, making it compatible with other popular automotive simulation platforms such as CARLA. These generated environments can provide a cost-effective and flexible alternative to physical testing, allowing researchers and developers to simulate different scenarios in a controlled and reproducible environment. Our approach has the potential to significantly improve the efficiency and scalability of automotive simulations by generating realistic environments for simulation purposes. The GitHub repository of this plugin is publicly available on https://github.com/fzi-forschungszentrum-informatik/eg4u.

Validation and verification of autonomous vehicles is still an unsolved problem. Although virtual approaches promise a cost efficient and reproducible solution, a most comprehensive and realistic representation of the real world traffic domain is required in order to make valuable statements about the performance of a highly automated driving (HAD) function. Models from different domain experts offer a repository of such representations. However, these models must be linked together for an extensive and uniform mapping of real world traffic domain for HAD performance assessment.Hereby, we propose the concept of a co-simulation architecture built upon the Robot Operating System (ROS) for both coupling and for integration of different domain expert models, immersion and stimulation of real pedestrians as well as AD systems into a common test system. This enables a unified way of generating ground truth for the performance assessment of multi-sensorial AD systems. We demonstrate the applicability of the ROS powered co-simulation by coupling behavior models in our mixed reality environment.