is a academia- and industry-oriented fork of In content and features, it largely overlaps with, and offers additional functionality to support driver’s training simulators and the development of advanced driver-assistance systems.

BeamNG developed a custom soft-body physics engine, that serves as a backbone for a sandbox vehicle simulator. is organized around the simulation, and provides BeamNGpy that supports automated generation and execution of test scenarios, a vehicle control interface, collection of simulation data for training procedural content generation, as well as validation and verification. It focused on ground-based road-vehicle simulation but also offers air and maritime simulation using mods from the gaming community (

The version offers features more appropriate for a research context than a gaming one ( because it provides driving simulation software, and virtual tests, for the development and testing of autonomous vehicles, ADAS and vehicle dynamics. That is possible thanks to our sensor suite which is commonly used in autonomous driving such as camera, LIDAR, ultrasonic, electrics, and IMU. Each sensor can be customize to meet the individual needs of different applications.

BeamNGpy is the official open-source Python interface for The library implements a scenario-based approach: In a script the user configures vehicles and defines the sensor setup. This facilitates data collection for learning based systems and allows validation & verification of autonomous driving software. was released under a mix of commercial and open-source licenses by BeamNG GmbH. Academic customers are eligible to apply for a free access to the full version via the following page. Regarding commercial and industrial licensing opportunities contact us on

Our Roadmap for includes many improvements in the next few years. These upgrades will further enhance autonomous vehicle development and testing as well as driver’s training by integrating industry standards (OpenDRIVE, OpenStreetMap, OpenScenario) or integration with different types of learning environments.

This documentation will be regularly updated so consider it as a living document.

Welcome and enjoy!



Page created: 29 December 2018, at 11:02
Last modified: 17 April 2021, at 17:12

Any further questions?

Join our discord