Sensor Modelling and Virtual Testing in Cloud Infrastructures

The complexity of autonomous driving (AD) functions and systems leads to an increased effort in verification and validation. Therefore, complex simulation environments for virtual validation of combined and integrated AD systems are needed.

Within the IAV HEAT (Hamburg Electric Autonomous Transportation) project, a driverless vehicle for regular operation in Hamburg public transport was developed. To support the validation of the AD-components, a multi-sensor-simulation using CarMaker was build up and integrated in a highly automated test environment, which is connected to IAV´s AD software development platform.

In the next step the generic composition of the simulation was shifted to a cloud infrastructure offering a higher scalability and efficiency of resources. This can lead to a significant increase of test coverage and test depth i.e., parameter variation of base scenarios. Containerised base configurations encapsulate tools which are connected through generic interfaces. This ensures a flexible usage of the infrastructure and high adaptability to changing conditions and requirements.

The IAV approach is to build up a virtual reproducible scalable testbench running in cloud infrastructures for validation of AD functions. As CarMaker is suitable for cloud simulation, it has been used in this setup, providing a powerful environment simulation.