An Agent Driven Approach for Test Specification Generation for Software Defined Vehicle Features

Automotive industry is currently experiencing a major transition towards Software-Defined Vehicles (SDV) to facilitate seamless new feature integration via over the air (OTA) update. SDVs integrate more than 50 + ECUs with more than 400+ software functions, leading to significant software complexity and a substantial need for comprehensive bench level testing to prevent software issues at vehicle level. To address these challenges, a robust and edge case focused test specification is essential, ensuring maximum requirement coverage across all scenarios while incorporating insights from past bugs and field issues. Currently, the automotive industry faces challenges in producing high-quality test specifications mainly focusing on functional testing (both positive and negative scenario), fault injection testing, ISO standards testing, network management testing which often exceed 2,000 + test cases for SDV features in parallel with maintaining consistency over time, ensuring maximum system-requirement coverage, and meeting project milestones. This paper introduces a new method to overcome above said problems. We provide a comparative study involving our proposed method with other methods to show the efficacy of the proposed method.