go back

Determination of relevant Coorperation Partners by Machine Learning

Pascal Lieser, "Determination of relevant Coorperation Partners by Machine Learning", TU Darmstadt, 2022.

Abstract

In this paper, the relevance of road users for a possible future interaction is investigated. For this purpose, the concept of interaction is first discussed. This is described from various scientific perspectives using different methods and terms. Subsequently, an overall definition for interaction in road traffic is explained. In this work, a data-driven approach is followed. Several datasets containing interactive traffic situations are available for the analysis. It is discussed which dataset fits best for the analysis and which traffic scenario will be addressed in more detail to be able to make a statement about interaction between road users. The process of interaction within the scenario will be related to the overall definition. A suitable Machine Learning approach will be chosen for the evaluation of the data. Software is implemented that interprets the dataset and quantitatively describes traffic situations of interest for analysis. Based on this, a Machine Learning model predicts which road users are candidates for interaction in a given traffic situation. The evaluation is performed in 3 different variants. Finally, the results are discussed and it is considered how approaches for further work could be based on them.



Download Bibtex file Per Mail Request

Search