Duration: 15:20

401 views

6

Driving simulators are valuable instruments to analyze driver behavior in specific traffic situations, e.g. near accidents. It can also be uses to analyze the perception of changes in the infrastructure, e.g. dynamic road signs, and the drivers’ reaction.

Recent studies analyzed whether personalized messages on dynamic road signs can increase the situational awareness of the driver without distracting him or her. In order to achieve realistic driving behavior and create an immersive environment, a realistic 3D road model and traffic simulation are required.

For the traffic simulation, Sumo is a good candidate as it is free to use and it can be manipulated with TraCI to create specific traffic situations. There are also some documentations how to connect Sumo to Unity, which is a widely used 3D game engine . However, the most common data source to create a road network in Sumo – OpenStreetMap – is not detailed and accurate enough for driving simulators. Therefore, we developed a different approach using open georeferenced data.

The German state North-Rhine Westphalia provides georeferenced LiDAR point clouds and satellite images as well as xyz-files on GEOportal.nrw. To create a terrain model, the xyz-files are used as a height map. We then import the terrain model into RoadRunner from MathWorks, which is a software for creating 3D scenes for simulation and testing of autonomous driving systems. Using the satellite images and LiDAR scans we manually model the road sections in the study area.

RoadRunner provides a sign editor to insert road signs into the scene as well as assets for traffic lights. The 3D model is then exported in the OpenDrive format and imported into Sumo using Netconvert. It is also exported as a Unity object and imported into Unity. During the simulation, we use TraCI to synchronize the positions of the driving simulator vehicle and the surrounding traffic between Sumo and Unity. Thus, the surrounding traffic reacts to the test subject and vice versa.