Autonomous driving encompasses a range of complex tasks such as perception, scene understanding, and decision-making. These tasks must be seamlessly integrated to achieve the overall goal of autonomous driving. The software architecture delineates the distribution of software functionalities to individual components and orchestrates their interactions within the Electronic Control Unit (ECU) for Advanced Driver Assistance Systems (ADAS). Ensuring the vehicle’s safe operation requires correct implementation of the software architecture and proper interaction among the various software components as defined within that architecture. The AUTomotive Open System ARchitecture (AUTOSAR) standard serves as a foundational architecture framework for automotive software, with its Adaptive Platform playing a pivotal role in facilitating the use of high-performance hardware and enabling the development of dynamic applications essential for autonomous driving. This thesis provides an overview over requirements for software integration testing through literature review and expert interviews and explores challenges and approaches in practice. The interviews are then evaluated using a grounded theory method in order to identify requirements, challenges and approaches. Subsequently, these identified requirements are analysed to assess their priority and impact on other requirements. This thesis emphasizes International Organization for Standardization (ISO) 26262 and Automotive Software Process Improvement and Capability dEtermination (ASPICE) as critical standards that define the scope, methods, and environments for testing, underscoring their role as key requirements in the domain of software integration testing. Another key finding is the critical need for robust data manipulation techniques for interface testing and the necessity for test automation for thorough and efficient testing. In order to implement test automation, a set of robust tools is needed which are compatible and can be implemented in an automation pipeline. The thesis identifies major challenges, including the integration of AUTOSAR Adaptive within the ADAS ECU due to tool limitations, and suggests utilizing AUTOSAR’s logging and calibration features as a potential solution. Interoperability issues among tools and the systematic definition of worst-case scenarios for resource testing are also addressed as challenges. Last, the late positioning of testing is a challenge, which can be tackled by the implementation of more agile development processes or a test-driven mentality. Future work will involve stakeholders in further evaluating the identified requirements and refining approaches. Promising directions include the development of a logging and calibration framework, the adoption of test-driven development, the implementation of code generators, simulation environments, and the exploration of tools for the AUTOSAR Adaptive Platform. This thesis lays the foundation by providing a comprehensive overview of the requirements, challenges and approaches in software integration testing. The identified requirements play an important role in evaluating and refining approaches to software integration testing in future work, with the aim of improving testing in the field of autonomous driving.