COSMOS

Development of a Model-Based Engineering Methodology for Software-Defined Mobility Systems - Modeling and Analysis of AI Functions on Heterogeneous Hardware Platforms for Software-Defined Mobility Systems

01/04/2026 to 31/03/2029

Digitalization in the automotive industry is transforming the way vehicles are built and defined: from hardware-first to software-first. This creates opportunities to expand the system context of vehicles to include traffic infrastructure and IoT devices, thereby driving the transition from software-defined vehicles to software-defined mobility (SDM). COSMOS aims to initiate and demonstrate European SDM-oriented application development by providing solutions and supporting the development of applications that encompass SDVs, road infrastructure, consumer devices, and cloud integration. To achieve this, the project will provide a model-centric design and operational methodology as well as best practices, complemented by various design tools and automation support. Applicability will be demonstrated using SDM stack components and applications.

The University of Tübingen focuses on the modeling and analysis of AI functions within the framework of a model-centric DevOps methodology for software-defined mobility (SDM). The goal is to analyze AI functions and derive requirements for the hardware architecture regarding performance, runtime behavior, and memory requirements. To this end, performance estimators for different compute resources (CPU, GPU, NPU) will be developed to enable an assessment of the entire hardware platform. These methods will be integrated into the digital twin of the overall system to evaluate implementation alternatives and provide feedback for an optimization strategy.