Highly automated driving remains a major challenge-despite billions in investments and promising pilot projects for L4/L5 in China and the U.S. However, the road to market readiness is still long. One of the biggest bottlenecks is system validation, which is highly resource-intensive-technically, organizationally, and financially. Moreover, current validation processes extend development cycles significantly, slowing down time-to-market. The ongoing shift from rule-based AD stacks to “Full-AI” or end-to-end AI-driven architectures holds promise, but is still in its early stages. Even with powerful AI approaches, covering a wide range of ODDs remains a key challenge. At the same time, early availability of high-quality training data in sufficient quantities remains crucial-regardless of the chosen stack architecture. Capgemini presents an approach that significantly accelerates the development and validation of automated driving systems. This approach combines a data-driven ecosystem featuring a data marketplace, an open toolchain, and a comprehensive service portfolio. A key focus is the efficient generation of Ground Truth-leveraging both real-world data and virtual scenarios, as well as the future integration of AI-generated datasets. The latest research on AI-based scenario generation will be explored through insights from the NXTAIM project. Another critical component is an innovative training and evaluation pipeline, based on findings from the research projects KI-Wissen and GAIA-X4KI. This pipeline enables automated identification of weaknesses in AI models and targeted improvements-currently a highly labor- and compute-intensive process. Finally, we want to demonstrate how the loop between data marketplace and model optimization can be closed: seamless access to high-quality real-world and simulation data unlocks cost reduction and accelerates validation. The result: more efficient, safer, and economically viable automated driving systems.

In this session, you will deep dive into:

• Cross-company economies of scale to enhance efficiency in AD development
• Secure, trustworthy AI to optimize validation processes
• Automation and global work distribution to reduce costs and shorten development cycles