A real-time multi-sensor fusion platform for automated driving application development

Advanced Driver Assistance Systems (ADAS) become standard and sometimes mandatory for vehicles, e.g. autonomous emergency braking. Future vehicles will include multiple ADAS that assist with safety-critical operations. For efficiency and effectiveness, these ADAS should share resources, information and functionalities. Additionally, ADAS performing safety-critical functionality require predictability for the execution of their processes. This paper presents the integration of a layer-based multi-sensor fusion and processing platform into a real-time system that also supports non-critical processes, i.e. a mixed-criticality system. A suitable system is selected, the behavior of the platform and its interfaces is described and tests are performed to validate the predictable behavior, by examining the difference in jitter and execution latency. The real-time layer-based platform is suitable for the development and testing of multiple integrated safety-critical ADAS.