Motivation

At IPEK we develop new approaches to the validation of automated and autonomous driving systems. With the trend towards end-to-end (E2E) architectures, the typed intermediate representations of modular architectures disappear — and with them, important anchor points for design-time assurance, while open operating environments leave a residual risk that no finite test campaign can eliminate. Assurance must therefore continue during operation — through runtime monitoring: the assessment, at runtime, of whether the output of the E2E driving function is trustworthy and, above all, safe.

Tasks

Systematic identification and classification of current runtime safety monitoring approaches for automated driving, based on a literature review following established systematic review methodology (e.g. defined search strings, databases, inclusion and exclusion criteria)
Development of a consistent taxonomy that orders monitoring approaches by the information they require from the monitored driving function
Classification of selected, publicly documented AD stacks within this taxonomy with respect to architectural approach, modularity, and existing assurance and monitoring concepts
Identification of research gaps and trends, particularly in the trade-off between performance and assurability of data-driven systems
Scientific write-up of the thesis (German or English)

Your profile

Studies in mechanical engineering, industrial engineering, computer science, electrical engineering or similar
Interest in autonomous driving, functional safety and current developments in AI
Basic knowledge of automated driving systems, AI/ML or vehicle architectures is an advantage
Analytical and structured way of working, and the ability to present complex technical content clearly and comprehensibly
Confident academic writing in English

We offer

Work on a current and scientifically relevant topic in the field of automated driving
Close technical supervision and methodological support in conducting a systematic survey
In-depth insight into modern AD architectures and assurance strategies
The opportunity to build solid expertise in autonomous driving
The option to make further use of the results (e.g. conference paper, journal survey, technical talk), depending on quality and interest

Earliest starting date: Immediately

Interested? Please submit your application with a CV and grade overview to anian.scheibel∂kit.edu

Starter-Literature: