Homologation of Driving Functions with Generated Test Scenarios

• Simulating critical situations with data from real test drives
• Deriving new test cases by modifying existing simulation scenarios
• Validating sensor models by comparing measurements from the real world with those from simulation

Task: Validating Driving Functions in SIL Simulation

Test scenarios have to be generated and used in software-in-the-loop simulation to validate driving functions.

Challenge: Proven Realism of Simulations

The validation of driving functions requires billions of test kilometers. Since real test drives are out of the question for time and cost considerations, software-in-the-loop simulation is used and executed faster than in real time.

The challenge for this type of simulation is to reliably prove that the tests accurately represent reality. For example, the scenarios based on test catalogs such as EuroNCAP very rarely represent situations that occur for a driving function in a real traffic situation.

Solution: Scenarios Based on Real Test Drives

For the above reasons, the scenario generation service from dSPACE and understand.ai transfers the measurement data from real test drives to simulation scenarios. This lets you set up a catalog of realistic scenarios. Situations that are rare in real life (edge and corner cases) can be safely reproduced in simulation.

By analyzing the measurement data before simulation, you can immediately identify these particularly relevant situations.

In addition, the logical simulation scenarios can be parameterized. This makes it possible to create new variants and therefore new test cases. You can also uncover new corner cases by adjusting the parameters, for example, for scenario-based testing.

Benefits of Scenario Generation with dSPACE and understand.ai

Since scenario generation is not based on a fixed sensor setup, different types of measurement data (raw data and preprocessed object lists) can be used. This means you can also use measurement data that was captured with old vehicle models, for example, eliminating the need for a measurement vehicle for scenario generation.

The process is automated to a very high degree, giving it the scalability required to generate a large quantity of scenarios in a very short time.

The dSPACE simulation environment provides the simulation scenarios, ensuring their seamless integration in the dSPACE validation solution chain (software- and hardware-in-the-loop).

Finally, the scenarios are also made available for OpenSCENARIO and OpenDRIVE, so you can also use them in other simulation environments. This enables Tier 1 suppliers to pass their test catalogs to an OEM with the driving function.