Safety Case Preparation for Road Autonomous Systems
Abstract
SynDAB enables simulation-driven safety testing and validation of autonomous systems. It supports safety case development and functional certification under ISO and PAS standards by simulating real-world corner cases with traceable requirements mapping.
Advantages
- Coverage of rare, safety-critical scenarios in simulation
- Conforms to ISO 17757, 22737, 21448 and PAS 1883
- Enables KPI-based performance benchmarking
- Facilitates safety case preparation and certification readiness
Fields of application
- Pre-certification testing of autonomous vehicles
- AI safety evaluation tools for OEMs and auditors
- Certification and risk assurance services
Background
Urban autonomous systems—such as service robots, last-mile delivery vehicles, and autonomous shuttles—must navigate highly unpredictable environments involving pedestrians, cyclists, traffic flow, and urban infrastructure. Ensuring the safety of such systems under a wide range of operational conditions is a core requirement of emerging standards like ISO 21448 (SOTIF) and PAS 1883. However, real-world testing is prohibitively complex, potentially unsafe, and unable to reproduce edge scenarios systematically. A simulation-based approach with traceable synthetic data offers a scalable solution to these challenges.
The invention originates from the Institute of Industrial Automation and Software Engineering (IAS) at the University of Stuttgart, led by Prof. Michael Weyrich (Institute Director) and Prof. Christof Ebert. Their research focuses on AI-based testing, safety validation, and certification readiness of autonomous and safety-critical systems operating in dynamic and human-centered environments.
This invention builds upon prior work in software testing, simulation-driven validation, and safety engineering for autonomous vehicles and robotics. It extends these concepts into urban driving scenarios, enabling scalable, traceable, and standards-aligned validation of AI-driven road systems.
Problem
Urban autonomous systems must be tested across a vast range of real-world scenarios involving traffic, pedestrians, infrastructure variability, and human unpredictability. These tests are difficult to replicate, time-consuming, and potentially dangerous, especially when evaluating system response to edge cases or system failures.
Validation according to modern safety standards such as ISO 21448 (SOTIF) requires reproducible and traceable results from scenario-based testing. Without a structured and risk-prioritized simulation framework, developers lack the tools needed to ensure reliable system behavior in complex urban environments.
Solution
The patented SynDAB approach supports simulation-based validation of autonomous systems in urban environments, including service robots, shuttle vehicles, and smart delivery systems. AI-generated traffic, pedestrian, and environmental conditions are used to generate a risk-prioritized set of test scenarios for system validation.
This test framework enables the creation of a Minimum Viable Test Set (MVTS) that reproduces corner cases such as unexpected pedestrian movement, occluded visibility, or sensor faults. System responses can be mapped against standards such as ISO 21448 (SOTIF), ISO 26262 (functional safety in vehicles), and PAS 1883, with full traceability.
The system can be integrated into digital twin platforms and supports simulation-based audits for regulatory submissions, providing transparency and safety assurance for deployment in public urban spaces.
The technology can be commercialized through licensing, offering a straightforward pathway for OEMs, Tier 1 suppliers, and validation tool providers to integrate SynDAB into their testing infrastructure. Additionally, the inventors’ team at the University of Stuttgart can offer cooperation or consulting projects to adapt the framework and algorithms to specific domains, including autonomous driving, last-mile delivery, and regulatory certification environments.
![A symbolic representation of SynDAB supporting urban safety validation: AI models for autonomous systems are tested through synthetic simulations of city environments with traffic, pedestrians, and infrastructure, enabling scenario-based risk analysis and continuous safety assurance.[Image: AI-created by TLB GmbH]](/fileadmin/user_upload/img/Technologieangebote/19_006_Driving.png "A symbolic representation of SynDAB supporting urban safety validation: AI models for autonomous systems are tested through synthetic simulations of city environments with traffic, pedestrians, and infrastructure, enabling scenario-based risk analysis and continuous safety assurance.[Image: AI-created by TLB GmbH]")
A symbolic representation of SynDAB supporting urban safety validation: AI models for autonomous systems are tested through synthetic simulations of city environments with traffic, pedestrians, and infrastructure, enabling scenario-based risk analysis and continuous safety assurance.[Image: AI-created by TLB GmbH]
Publication and links
Ebert, C. et al. (2023). AI-Based Testing for Autonomous Vehicles. ResearchGate link
IAS (2022). Testing Software Systems. University of Stuttgart. Link
Fickinger, M. et al. (2023). Simulation-based Validation for Construction Robotics. Automatisierungstechnik, De Gruyter. Link