How FDTech Utilizes CarMaker for Developing HAD Functions
FDTech GmbH and its leading engineers are experts in developing autonomous driving functions and everything else that goes with it. This includes not only the implementation, but also requirements definition and elicitation, as well as the whole validation process. For all these engineering steps along the V Model, FDTech has build up a so-called Digital Twin. The simulation tool IPG CarMaker supports the FDTech Digital Twin at various steps of the engineering process.
Regarding the requirements elicitation and system design, CarMaker is used to test different prototypical implementations of driving functions. E.g., for a longitudinal PI controller several parameter constellations are tried out within the configuration of the functionality to check if certain situations are still drivable. Parameters for the final definition of the requirement are then chosen according to the results.
For the validation of implemented driving functions, CarMaker is used within the SIL environment for running thousands of different test cases which have been derived through scenario-based testing approaches. These tests run permanently within the CI/CT cycle, where each code adjustment is followed up by an automatic integration in the whole software. DevOps to its perfection.
Finally, CarMaker is used by FDTech to train AI-based driving functions for situation interpretation and decision making. An example for such a function is the trigger for a lane change. In this case the simulation is used as a training playground. The AI is taking actions and learns the best behavior thanks to a reward function that considers certain relevant KPIs like Time to Collision. Reinforcement Learning enabled by IPG CarMaker.