Recording-Pipeline
Records driving video and driving data of a simulator run on a shared time base.
What & why
Video-based display studies need driving video and driving data on a shared time base. To align them by hand is error-prone. The Recording-Pipeline couples both at the moment of recording: The StudyControlApp starts and stops the recording over MQTT, and the OBSController turns the command into OBS Studio calls. EcoSimLabPy writes a recording flag into the telemetry log; the flag marks the ticks that are also in the video. Two parts of the chain are not public: The OBSController is in an internal repository, and the final FFmpeg step to Full HD is manual work without a script. The chain is not bound to BeamNG.tech — for a different simulator, write the recording flag into its data stream.
The recorded runs feed both the VideoSimLab studies (as stimuli for the Cockpit-Display-Replay) and the planned demo recordings. For real OBD/CAN vehicle data, CanZe is the documented starting point; the high-frequency logging, however, has to be solved there by the user.
Because there is no repository of its own, no license is set. The tool therefore stays experimental. The guide describes the four steps of the chain.
Cite
- Jan Heidinger, Leonardt Wagner, Christopher Spölmink, Thomas Franke (2026).VideoSimLab: Developing a Video-Based Toolchain for Vehicle Display Evaluation StudiesAutomotiveUI '26 (18th Intl. Conf. on Automotive User Interfaces)10.1145/3828157.3828768