Quality assurance of AI – the dangers lurk around the corners

Blog | Development

The dangers are hiding behind the corners and QualityMinds are investigating the so-called corner cases. Currently, we are celebrating our first successes in a project involving quality assurance of artificial intelligence. And we are happy to learn a lot in the process!

The quality of AI systems is of crucial importance, especially in the complex area of autonomous driving. Here, even in the most unusual situations, algorithms, traffic law and sensor technology must interact perfectly in order to make the right decisions. Where human lives are at risk, it is particularly necessary to keep the error rate as low as possible. This is a challenge we take up with pleasure.

Since July 2019, QualityMinds has been working closely together with 27 partners from the automotive industry and the field of research (OEMs, suppliers, AI start-ups and research institutes) within the framework of the “AI Assurance” project of the German Federal Ministry for Economic Affairs and Climate Action. The framework is funded with 41 million Euros, in order to guarantee the assurance and release of AI systems in autonomous vehicles through a stringent and verifiable chain of argumentation. This requires, among others, the training and testing of algorithms with data that represent as broad a spectrum of situations as possible. In return, it requires the synthetic generation of the data.
QualityMinds is leading a project sector dedicated to the development of a methodology for the identification of so-called corner cases (i.e. exceptional situations) and thus their consideration for the quality of the training and test data. We remain curious to see what else will happen in this industry and celebrate our first successes.

Thanks to our experiences with the aforementioned project as well as other consulting tasks, internal development and research projects, we are constantly evolving and broadening the portfolio dedicated to AI testing. The testing world and its paradigms will undergo significant change in the future due to the increasing use of artificial intelligence. We are strongly committed to learning more about these developments and look forward to discussions and exchange of experiences on this topic. We are also looking forward to consulting on quality assurance of AI in further projects.

If you want to find out more about the project and people involved in it, read our article in the German Testing Magazin. Have a look at the latest product from our Innovation Lab for the testing of AI in autonomous vehicles or visit our Autonomous Driving Testing page.


Autonomus Driving

written by

Veronika Sturm