Fully automated driving functions are a new dimension of driving - and thus, a new dimension in safeguarding and approving these driving functions.
Current methods and processes for safeguarding and approving are designed for assistance driving functions. However, for fully automated driving, many new challenges arise for the release processes, for the validation of the complete vehicle and in considering regulatory requirements.
Extending current assurance and approval processes
In a current project, the AQI is identifying potential gaps that would occur if the current assurance and approval processes were applied to fully automated driving functions. As a result, proposals to close these gaps are made. In addition to legal requirements, approval criteria, homologation requirements and simulation tool requirements are also being considered. The aim is to extend the safeguarding and approval processes to include fully automated driving in such a way that the requirements of customers, legislators and the industry are fulfilled.
Safeguarding and approving of AI software
In order to prove in a liability case that the state of the art was followed for safeguarding and approval, quality processes, documentation and verification management must be adapted to the complex characteristics of AI algorithms. Accordingly, the AQI is developing concrete solutions that will be integrated into future standardization work. For example, a new process group for A-SPICE has been developed that can be used for AI/ML software. Furthermore, the AQI is working on an FMEA for AI/ML-based applications.