From Traffic Sensor Data To Semantic Traffic Descriptions: The Test Area Autonomous Driving Baden-Württemberg Dataset (TAF-BW Dataset)

Abstract

The validation and verification (V&V) of highly automated driving depends on reference data in the different steps of the development process. While there exist a variety of datasets to optimize and benchmark perception algorithms, large trajectory datasets with complex interactions between traffic participants are quite rare. Such data offers the opportunity to derive virtual traffic scenarios and road user behaviour models that can be used to evaluate in-vehicle perception systems as well as decision making modules. We present the Test Area Autonomous Driving BadenWürttemberg (TAF-BW) Dataset that provides globally referenced road user trajectories extended with context information like traffic light signals and high definition map data recorded in the test area. We point out an exemplary application that becomes possible with the dataset: building semantic scene models that can be used for criticality assessment in traffic scenes. We conclude with a short outlook on future extensions.