Autonomous Driving Simulation

Autonomous driving is the future of individual mobility and all major manufacturers are working on fully autonomous vehicles. While there are robust and good solutions for the individual problems in autonomous driving, the main challenge lies in their integration. Altogether, an autonomous vehicle’s software is the biggest problem. Therefore, the key in self-driving vehicles is about getting the software right. Today, autonomous driving is both a vision and a reality, at least on SAE level 2 and most recently also level 3 on the SAE driving automation scale: https://www.sae.org/standards/content/j3016_202104/.


For more information about building autonomous driving systems, see our course AVAI (autonomous vehicles and artificial intelligence) or the book https://link.springer.com/book/10.1007/978-3-031-01805-3 (accessible with a RUB IP).

Carla, a driving simulator based on Unreal (source: unrealengine.com)

The goal of this project is to compare two open-source autonomous-driving platforms upon scenarios and maps you create. The latter should be related to RUB, ideally the campus. You should investigate ways of efficiently mapping the campus and build driving systems upon the autonomous-driving platforms.

In this study project you will continue work already done by past student groups and evolve the established foundation.

Further information on the current state of the project, can be found in our technical report on researchgate.

Obtaining RoadRunner License

To obtain a RoadRunner license, write an email request to IT-SERVICES using the template below.

Subject: Anfrage zur Lizenz für die RoadRunner-Produktfamilie

Sehr geehrte Damen und Herren,

mein Name ist [Name], und ich bin Student der Angewandten Informatik an der Ruhr-Universität Bochum. Im Rahmen des Studienprojektes „Autonomous Driving Simulation Case Study“, welches von Prof. Dr. Thorsten Berger geleitet wird, benötige ich eine Lizenz für die RoadRunner-Produktfamilie (RoadRunner, RoadRunner Scenario, RoadRunner-Asset-Bibliothek, RoadRunner Scene Builder).

Ich bedanke mich im Voraus und freue mich auf Ihre Rückmeldung.

Mit freundlichen Grüßen

[Name], (Matr. Nr.: …)

Note: The license will be assigned to the e-mail address provided in the request within 1-2 days.

Installing the License

  • Create a MathWorks-Account by going through the registration process (use your University email address).
  • Navigate to the MathWorks License Center. If you aren’t already logged in, log in with your existing MathWorks account to access the License Center.
  • Select the RoadRunner license by clicking on the row labeled RoadRunner (Individual). After the selection, you should see the RoadRunner, AssetLibrary, Scenario, and Scene Builder under the tab “Manage Products”.
  • Download latest available version of RoadRunner.
  • On the tab with the products listed, move over to the tab “Install and activate” and click on the blue button labeled “Activate a Computer”
  • Fill out the box titled “Manually Activate Software on a Computer” and then click on continue.
  • Download your license file and store it on your computer.
  • Double-click on the Shortcut to launch the program.
  • Provide the license path in the window that opens up on first launch.

Carla Python API

The Python API in CARLA serves as an interface between the CARLA simulation server and user scripts, enabling users to create, control and manipulate autonomous driving scenarios in real-time. Through the API, users can control vehicles, pedestrians, sensors and various environmental factors. The core parts of the API are:

  • Vehicle Control: The API allows users to spawn and manage vehicles within the simulation. Users can control essential driving parameters, such as throttle, brake, steering, and gear.
  • Sensor Simulation: CARLA supports a variety of sensors, including cameras, LiDAR, RADAR, and GPS. The Python API enables users to attach these sensors to vehicles, configure their parameters, and collect real-time data for perception and analysis.
  • Scenario Management: With the Python API, users can define complex driving scenarios involving interactions between multiple actors, such as vehicles, pedestrians, and traffic signals.
  • Environmental Control: The Python API allows for dynamic manipulation of environmental conditions within the simulation. Users can adjust weather (rain, fog, cloudiness), lighting conditions, and road friction in real-time to test how autonomous systems perform under various scenarios.
  • Data Collection and Analysis: The API supports logging and data collection from both vehicles and sensors. Users can programmatically retrieve this data for analysis, allowing the simulation to serve as a valuable resource for training machine learning models or evaluating vehicle performance.

For more information, refer to the official CARLA Python API tutorial.

Troubleshooting

Windows Installation of CARLA
It is nearly impossible to build and run the full version of CARLA under an up-to-date Windows 11 OS, directly nor via Virtual Box. We highly recommend to use the Ubuntu versions mentioned in the manual, installed natively on the hardware.

Missing Manifest Error in Blender
Depending on how you try to import the city data into Blender, you may run into this error. The cause is that the requirements for add-ons seem to have changed from Blender 4 onward. Therefore, you will not be able to install many add-ons in the latest version of Blender unless they have been maintained. While it is technically possible to update the add-ons yourself, you can avoid this issue by using the latest version of Blender 3. Any more modern features introduced in Blender 4 are not required for our use-case.

KeyError: ‘Geometry’ from CityGML or CityJSON File
Some of the available tools make assumptions about some fields definitely being set. Unfortunately, these assumptions are violated by many of the files that can be found online. Thus, we recommended using CityGML2OBJ 2.0. We found it can cope with such files, unlike other tools to convert or directly import the data.

Useful links

References

Sollmann et al., A Tutorial on Modeling Urban Environments for Autonomous Driving Simulations, 2025, https://www.researchgate.net/publication/392591339_A_Tutorial_on_Modeling_Urban_Environments_for_Autonomous_Driving_Simulations