Automotive Industry Insights

How does automated driving change the release and homologation process?

Parallel to this shift to data-driven development, as we have established here, the release and homologation of automated driving is also changing. In the past, homologation was the last step during or after development. Random samples were made on vehicles to test overall system behavior. Given the complexity of new systems, this is no longer sufficient. The residual risk of such a highly automated system cannot be determined at the overall system level. In the future, the development process and the (mostly simulative) methods used there will be the basis for the homologation of automated driving functions.

The development of new data-driven development processes should thus also aim to achieve homologation as efficiently and safely as possible, and to fulfill compliance requirements. This means by implication that the requirements of the relevant current and future safety standards must be met.

As such, automated-driving releases will be based on two pillars, largely implemented by simulation: firstly, verification, in which all measures to prevent known risks are checked, and secondly, validation, in which the system is pushed to its limits in relevant use cases to identify unknown risks. Both pillars combine simulation with vehicle testing.

Through combining the available methods of HIL and SIL safeguards, a statement can be made about overall performance and residual risk. The methods will range from classic HIL tests to highly scalable cloud-based simulation solutions that can check an almost infinite number of possible combinations of scenarios. Tool manufacturers must make their products cloud-capable and more flexible than ever before to be able to address the higher number of iterations between verification and validation and the system design. Offerings such as SaaS (safety as a service) might play a role in the future.

The changes and transformations have led as never before to the need to standardize and further explore fundamentals. The complexity arising from software-centric vehicles, the challenges of autonomous driving, and data-driven development can only be managed if standards are applied wherever possible. To make this visible and to put current standards and current research into this context, this needs to be analyzed globally.