OEM&Lieferant 1/2025

67 IT and Automotive Direct Injection of Raw Sensor Data into ECUs Reproducibly test and validate ADAS and AD functions in critical scenarios By Astrid Schmidt, Communication Manager Press & PR, Henning Kemper, Senior Specialist Editor and Florian Weindel, Product Manager Hardware, IPG Automotive Advanced driver assistance systems (ADAS) and automated driving (AD) functions have become an integral part of vehicle development as automated driving technology has gained ground. These systems depend on complex and extensive sensor configurations within vehicles and to operate reliably, comprehensive testing across various traffic scenarios and ambient conditions is essential. However, relying solely on real-world testing is neither cost-effective nor feasible due to the massive amounts of data needed and the challenges in replicating specific ambient conditions. These functions must additionally be adapted to different target markets, each with its own unique traffic and ambient conditions, all of which require tailored ECU software testing. Simulation presents an ideal solution to this challenge: Using a simulation environment, it is possible to create and test specific scenarios virtually. The CarMaker product family from IPG Automotive offers robust simulation solutions, modeling all essential elements such as vehicles, roads, road users, and objects. The visualization tool Movie NX adds to this by providing physically realistic simulations of cameras, along with detailed radar and lidar sensor models. This includes the ability to simulate weather and light effects, as well as other environmental conditions across various times of day and seasons. To support development even further, IPG Automotive developed the SensInject hardware solution, which serves as an interface between real ECUs and virtual cameras, enabling the direct injection of camera sensor image data into ECUs. A high-performance FPGA ensures that the available image data is accurately aligned with the ECU‘s camera interface, allowing for communication with the interface and the enrichment of image data with sensor-specific embedded data. The platform also enables targeted manipulation of the data, offering flexible options for failure injection. Both real-time simulated raw sensor data in closed-loop testing and prerecorded real-world data using replay methods can be deployed in this context, ensuring that the sensor emulation quality meets the necessary standards for use with series ECUs without the need for specialized software. Flexible and easily adaptable to customer needs SensInject is designed to be highly adaptable to a wide range of customer-specific use cases. Its modular hardware architecture allows for easy integration, with modules that can be replaced or added with minimal effort. It also supports established sensor interfaces, such as GMSL2, MIPI CSI-2 and FPDL4, making integration into respective test systems straightforward. As a result, the SensInject platform is already geared for future developments and the evolving demands of camera-based system testing. Safety-critical system validation under real-world conditions SensInject is designed for both closedloop and open-loop testing with simulated sensor data as well as for replay of recorded real-world data. This allows for deterministic, reproducible functionality tests, which are crucial for validating safety-critical systems under realistic conditions. Additionally, the platform supports fast and precise robustness tests and failure injections. Hence, specialists across various fields benefit from SensInject: whether for validating perception modules or conducting full system validation of ADAS and AD functions. For all camera sensor-based development and test processes, SensInject offers significant advantages, from versatile possibilities for application to direct sensor data injection. Its modular design provides high flexibility, contributing to enhanced test coverage, quicker validation of assistance systems and AD functions, and ensuring reliable system performance. IPG Automotive www.ipg-automotive.com  Image: © IPG Automotive The set-up shows a SensInject platform (left), an Xpack4 real-time computer (middle) and an electronic control unit (right) which is fed video data from the simulation by SensInject.

RkJQdWJsaXNoZXIy MjUzMzQ=