Researchers' Req2Road AI pipeline automates 89% of software-defined vehicle testing
The system uses LLMs and VLMs to turn natural language requirements into executable vehicle tests.
A research team from multiple universities built Req2Road, a GenAI pipeline for Software-Defined Vehicles (SDVs). It uses Large Language Models (LLMs) and Vision-Language Models (VLMs) to interpret requirements and automatically generate Gherkin test scenarios. The system then converts these into runnable scripts using the Vehicle Signal Specification (VSS) standard. In an evaluation, it successfully transformed 89% (32 of 36) requirements for a Child Presence Detection System into executable tests for virtual and real-vehicle execution.
Why It Matters
This dramatically accelerates automotive software validation, a major bottleneck in developing connected and autonomous vehicles.