ADSO
The growing complexity of automotive software, fueled by AI advancements, has intensified safety challenges as vehicle automation increases. Traditional verification and validation methods, though vital, are no longer sufficient. A shift toward the DevOps paradigm enables continuous monitoring, feedback, and updates after deployment, ensuring ongoing improvement of safety-critical systems throughout a vehicle’s lifetime. Unlike non-critical systems, implementing DevOps in safety-critical domains poses significant conceptual and technical challenges. This project aims to address these issues and establish a reliable DevOps framework for automotive software safety and continuous evolution.
Projectdescription
As digitalization drives automotive innovation, the rise of highly automated driving creates a critical challenge: ensuring the safety and certifiability of vehicle software that requires continuous updates after deployment. The MANNHEIM–AutoDevSafeOps (ADSO) research network was established to address this gap by creating a holistic framework for secure, certifiable, and modular runtime updates for safety-critical systems. At its core, ADSO introduced a DevOps-based lifecycle, the "ADSO process," which integrates continuous development, deployment, and validation into a unified workflow. This is embedded within the Reference Lifecycle Process (RLP), a V-model framework that ensures all activities comply with functional safety standards like ISO 26262 through continuous feedback loops and an integrated safety assessment process. To handle the "Open World" problem of unpredictable scenarios in Level 3+ automation, the project utilized advanced virtual testing and simulation to guarantee ongoing safety. ADSO’s work provides a foundational methodology for the future of autonomous driving, enabling secure updates and continuous improvement while maintaining Germany’s leadership in the automotive industry.
Focus points/areas
-
AutoDevSafeOps
-
Computer Vision
-
YOLO
-
CNN
-
Project duration
01.11.2023 - 30.09.2025 -
Areas
AutoDevSafeOps, Computer Vision, YOLO, CNN -
Financing
Funding agency: Federal Ministry of Education and Research (BMBF)
Methodology
Standardisierte Befragung von Lehrkräften
Die Datenerhebung erfolgt mittels Fragebogen unter Fachkonferenzvorsitzenden, um deren Wahrnehmungen, Einstellungen und Erwartungen systematisch zu erfassen.
Operationalisierung zentraler Implementationsdimensionen
Die Wahrnehmung der Reform wird anhand etablierter Dimensionen wie Vorteil, Passung, Komplexität und Umsetzbarkeit differenziert analysiert.
Faktoren- und Clusteranalyse zur Typenbildung
Durch Hauptkomponenten- und Clusteranalysen werden unterschiedliche Lehrkräftetypen identifiziert, die sich in ihren fachlichen und didaktischen Orientierungen unterscheiden.
WP 4: Researches methods to secure updateable automotive systems and enable efficient, incremental validation and certification.
WP 5: Implements the necessary middleware, hardware, and communication functionalities to support runtime updates and monitoring.
WP 6: Tests and demonstrates the project's integrated technologies and processes within practical, real-world use cases.
