Advanced driver-assistance systems (ADAS) are one of the key technologies for future innovations in cars and commercial vehicles. Increasingly, these systems take over safety-related functions, e.g., autonomous braking and autonomous steering. Typically, the systems are based on radar sensors or cameras continuously monitoring the vehicle environment. These sensor technologies cannot provide completely reliable detection rates of objects. Correspondingly, the quality assurance for advanced driver-assistance systems is of high importance. In this keynote an approach for automatically testing camera-based driver-assistance systems will be presented. Systematic test design and test scenario generation is based on the classification-tree method, environment simulation and image generation are supported using vehicle dynamics and virtual test drive simulations, such as veDyna or CarMaker, and test execution and test evaluation are performed using the test environment MESSINA. The proposed test solution enables functional testing of ADAS features in different testing phases, e.g., software in the loop for testing functional models in early development phases, hardware in the loop for testing the electronic control units (ECU) with the integrated software, as well as back-to-back tests between functional models and ECUs. Additionally, the high degree of automation enables efficient regression testing.
展开▼
