The development of automotive electronic systems can be divided into three main test phases: function testing; ECU testing, including perception and fusion; and ECU domain and network testing. Since each test phase focuses on different goals, the test system requirements differ as well. The dSPACE HIL systems can be adapted to each phase to provide maximum support.
Function testing aims at testing single or few functions, including their interaction with ECU basic software. Since this kind of testing is typically an integral part of function development rather than final release testing, its focus lies on experimental usage, without systematic test automation. Hardware-in-the-loop (HIL) test systems for function testing therefore require versatile bus and network interfaces, restbus simulation on the behavior and protocol levels, or design-for-test interfaces (e.g., XCP). If you want to test functions for autonomous driving, the test system must be able to generate sensor data on the object list level. These types of tests are often performed by the function developers themselves. Therefore, the test systems must fit on a developer’s desk.
Typical applications for ECU tests range from software integration testing over to fail-safe and fail-operational testing, to ECU release testing. For ADAS and AD ECUs, this can even include perception and fusion algorithms also have to be verified. Testing is therefore highly automated and based on ECU requirements to cover the high number of variants. Scenario-based testing is also useful to address the many test variants, particularly in combination with automatic scenario generation.
Test systems for ECU testing require scalable I/O and bus/network interfaces, restbus simulation on the behavior and protocol level, and extensive manipulation and test options for buses and networks. Scenarios for autonomous driving are often tested using raw sensor data. These tasks require comprehensive plant models and simulation performance that can be increased accordingly. Ideally, the test system functions can be adapted to changing project requirements.
ECU domain and network testing is characterized by end-to-end testing of distributed functions. This includes fail-safe and fail-operational tests as well as vehicle network testing, such as sleep or wake-up tests. The systems often run highly automated test programs around the clock, made possible by comprehensive test management systems.
The need for an extensive range of I/O channels as well as bus and network interfaces is characteristic for these kinds of test systems. To handle complex models and scenarios, they also require strong simulation performance and multiprocessor capabilities. The systems process a multitude of plant models, from powertrain to air conditioning to window lifter models, to name but a few. To test systems for autonomous driving, sensor signals must be simulated on either the object list or raw data level, depending on the system architecture.
Subscribe to our newsletters, or manage or delete your subscriptions