Automated System Test Generation for Software-Defined Vehicle Validation

The introduction of the System Test Generation (STG) Agent addresses the increasing complexity of the automotive data ecosystem, where the transition toward software-defined vehicles (SDV) requires the management of massive volumes of system specifications. This technology automates a traditionally manual, work-intensive stage of the validation cycle by translating R&D system requirements into traceable test protocols for automotive components and software.

Technical Mechanism and Integration
The STG Agent was developed in collaboration with the Amazon Web Services (AWS) Generative AI Innovation Center. The architecture utilizes Amazon Nova foundation models and Amazon Bedrock Knowledge Bases, integrated via the Strands Agents framework. This technical stack allows the tool to parse complex engineering data and identify the specific expected behaviors implied by human-defined system requirements.

In the standard development workflow at Marelli, engineers first translate customer needs into technical system requirements. The STG Agent then analyzes these requirements to automatically generate corresponding test cases. This automation ensures that every functional requirement is mapped to a validation step, maintaining strict consistency and traceability without the lead times associated with manual script authoring.

Impact on Validation Cycles and Reliability
As vehicle platforms become more reliant on frequent software updates and complex electronic control units, the volume of engineering data grows exponentially. The STG Agent serves as a bridge in the digital supply chain, ensuring that product features are validated against customer requirements with higher precision.

By reducing the time required for the generation of test cases, the tool enables vehicle manufacturers to accelerate product development cycles. The primary technical benefit is the achievement of stronger alignment between theoretical system requirements and actual validated product behaviors. Furthermore, the solution is designed for seamless integration with existing requirement management tools, ensuring compatibility with established automotive engineering standards and workflows.

Application in Automotive Engineering
The application of the STG Agent is specifically targeted at high-complexity systems where manual validation is prone to bottlenecks. By providing clear and structured test cases, the tool assists Marelli engineers in confirming that features behave as intended before they reach the production stage. This systematic approach to validation supports the delivery of new vehicle functionalities with a higher degree of reliability, particularly in the context of the rapid iteration cycles required for modern automotive software.

Edited by Evgeny Churilov, Induportals Media - Adapted by AI.

www.marelli.com