TIER IV, the pioneering force behind the world's first open-source software for autonomous driving, is proud to announce its selection for a major Japanese government project aimed at expanding deployments of autonomous driving services to address local mobility challenges. Through this initiative, TIER IV aims to advance autonomous driving standards for Level 4+ and deliver an open dataset that enables the development of highly transparent and traceable end-to-end AI.

Under this project, TIER IV will work closely with consortium members to develop industry-leading standards for autonomous driving, with automotive manufacturers playing a central role. This will facilitate the deployment of new autonomous driving services across the country. The consortium will also explore global market opportunities, especially in regions facing similar mobility and urban development challenges.

In addition, the consortium will develop an open dataset by integrating real-world data with data created by generative AI. This will enable the creation of simulation environments that replicate diverse traffic scenarios, essential for evaluating safety and performance.

In April 2025, TIER IV launched a collaboration with Carnegie Mellon University to develop a hybrid architecture that combines data-centric AI approaches with best practices in robotics, aiming to realize a next-generation Level 4+ autonomy concept. By leveraging the insights and outcomes from this collaboration, TIER IV will advance the development of autonomous driving standards for Level 4+, while creating and providing an open dataset designed to ensure transparency and traceability in the decision-making processes of embedded AI.

Focus areas

Efforts to establish autonomous driving standards for Level 4+ include:

• Integration of multiple end-to-end generative AI models into autonomous driving systems, with scenario-based evaluations used to compare their characteristics.
• Development of a framework for evaluating and validating generative AI models, along with a roadmap for real-world deployment.
• Application of safety evaluation techniques to robotaxi services, drawing on findings from proof-of-concept tests to support safe service design.

The creation and delivery of an open dataset that enables the development of highly transparent and traceable end-to-end AI involves:

• Collection of driving data using Level 4+ vehicles in and around Tokyo.
• Development of a large, diverse dataset by combining real-world and synthetic data to evaluate autonomous driving systems in complex environments.
• Generation of varied driving scenarios using different generative AI architectures, reflecting a wide range of behaviors, traffic participants, and environmental conditions.
• Comparison of synthetic output with real-world data to address generative AI challenges such as errors and hallucinations. Performance will be validated in real systems to confirm model utility, identify weaknesses, and support continuous improvement through MLOps pipelines.
• Delivery of a large-scale dataset through the evaluation of generative AI models in partnership with automotive manufacturers and the definition of synthetic data requirements.

The Japanese government's Mobility DX Strategy aims to establish profitable autonomous taxi services in the near term. TIER IV's participation in this project will not only contribute to this goal, but also lay the groundwork for applying autonomous driving technologies to passenger vehicles, further strengthening the global competitiveness of Japan's automotive industry.