A comprehensive test strategy using sensor data is required to ensure the proper operation of computer vision-based perception and sensor fusion algorithms. This presentation highlights hardware-based data reprocessing as an essential method for validating perception. Major adaptation of recorded sensor data is needed during reprocessing to meet frequent software and hardware updates, board network changes and requirements like E2E security protection along the development process. The presentation also explains how to validate the proper function of replay systems, which can have a tremendous impact on test quality and development costs. Finally, a brief outlook on AI-supported testing is presented.

• When to pick hardware-based data reprocessing and why it should be made an integral part of the strategy for testing perception and data fusion ECUs
• The major advantages of adapting recorded data during reprocessing and why this can be a challenging task
• Challenges of using sensor interfaces in reprocessing testing and how to solve them
• How to prove the proper function of a data reprocessing system to further increase test quality
• How GenAI-based methods can help to reduce the costs of reprocessing real sensor data