Brief Abstract
To ensure safe and comfortable automated driving, even the most complex traffic scenarios must be interpreted correctly. Since the static parts of the world rarely change, maps can be used to model them and their interpretation. This allows the interpretation to be performed offline and possibly even with human assistance.
This lecture will introduce Lanelet2, a map handling framework for the research and development of automated driving functions and autonomous vehicles. Lanelet2’s core idea is bottom-up design that not only enables reproducible mapping. The framework also connects the abstract map layers with the physical elements from which they naturally emerge. By unifying the needs of all driving functions, Lanelet2 ensures a consistent view of the world. In addition to the map framework, the lecture will show how Lanelet2 maps can be created using data from onboard sensors and/or aerial imagery. Finally, we will discuss currently developments around Lanelet2, such as an extension for deep learning.
- Date: 07.12.2023
- Location: Online Webinar
- Speaker: Jan-Hendrik Pauls, KIT
Presenter Bio
Jan-Hendrik Pauls received his B.Sc. in Engineering Cybernetics and his M.Sc. in Computer Science from the University of Stuttgart, Germany. He leads the research group for mapping and localization at the Institute of Measurement and Control Systems at the Karlsruhe Institute of Technology (KIT), where he also pursues his PhD. Within the field of automated driving, his research focus lies on perceiving the static world, capturing it in meaningful representations like HAD maps, highly accurate 6D localization within such maps, and their temporal verification.
