Scientific publications

It is essential for safety-critical applications of deep neural networks to determine when new inputs are significantly different from the training distribution. This paper explores this out-of-distribution (OOD) detection problem for image classification using clusters of semantically similar embeddings of the training data and exploit the differences in distance relationships to these clusters between in- and out-of-distribution data.

Model compression is widely employed to deploy convolutional neural networks on devices with limited computational resources or power limitations. For high stakes applications, it is, however, important that compression techniques do not impair the safety of the system. This paper investigates the changes introduced by three compression methods that go beyond the test accuracy.

A serious problem in image classification is that a trained model might perform well for input data that originates from the same distribution as the data available for model training but performs much worse for out-of-distribution (OOD) samples. This paper proposes a first-of-its-kind OOD detection architecture named OODformer that leverages the contextualization capabilities of the transformer.

The quality of semantic image segmentation models can be affected by external factors such as weather or daytime. Those factors can lead to safety-critical mistakes. In this work, we propose a systematic approach to detect and alleviate such weaknesses of semantic segmentation models.

Automated cars have to take correct decisions in complex situations. For this, the understanding of a vehicle system’s own capabilities and the environmental context is crucial. We introduce our approach of enhancing the system awareness of vehicles to handle changes gracefully, while optimizing the overall performance.

This paper describes the application of machine learning techniques and an assurance case for a safety-relevant chassis control system. The paper highlights how the choice of machine learning approach supports the assurance case, especially regarding the explainability of the algorithm and its robustness.

In this position paper “Towards Collaborative Perception in Automated Driving: Combining Vehicle and Infrastructure Perspectives” the authors propose a framework to merge the vehicle and the infrastructure perspective enabling collaborative perception and thus to enhance the dependability of the environment perception of automated vehicles in complex scenarios.