Towards a Virtual Continuous Integration Platform for Advanced Driving Assistance Systems

Abstract

Recent technological progress in computational engineering and systems design will enable the vision of autonomous driving coming true anytime soon. Functional but particularly also qualitative aspects of automotive functions are therefore gaining in importance more than ever before. This is due to the growing complexity of modern vehicles that gradually evolve into cyber-physical systems giving rise to the increasingly ambitious challenge of reliably validating the non-/functional integration of all their inherent subsystems. Thus, whereas traditional approaches to component and system testing are becoming more and more inappropriate for costs and general viability reasons, simulation-based methodologies offer an adequate solution due to their commonly scalable and generic nature. However, this is only true given a sufficiently high fidelity of the applied simulation models and a straightforward-in-use yet powerful-in-service evaluation platform with flexible support for execution semantics nesting, co-simulator coupling, and interfacing downstream tools with monitoring and visualization capabilities. In this regard, we introduce our concept of a continuous integration platform allowing for virtually prototyping technical systems of any kind that is applicable at any stage of the development process thanks to arbitrary levels of abstraction and wide-range tooling compatibility. This platform is based on the approved FERAL simulation framework at its core combined with versatile architectural components that are adaptable for domain-specific and cross-domain use cases. We focus this work on Advanced Driving Assistance Systems (ADAS) functions and showcase the end-user operation of the instantiated platform from the configuration of traffic scenarios over adjusting the functional logic and parameter values up to the visual validation of simulation results.