This paper presents a novel general evaluation framework for optical in-vehicle occupant classification systems. The proposed solution includes a guideline for systematic evaluation of possibly influencing parameters for vision based in-vehicle applications, definition of key parameters, evaluation and analysis of arising conditions. Subsequently, a manageable test matrix and suitable test procedures are derived, using a DOE1 approach. Furthermore, the functionality and underlying intent of an image acquisition and injection hardware for automotive HIL2 validations is discussed. The functionality of the presented hard- and software toolset not only includes capturing image data, but also the surveillance of the algorithm performance by monitoring CAN signals, and re-injection of the captured data in offline tests in order to be able to perform rapid and efficient software evaluations (i.e. algorithm updates, software changes).
Verification of optical in-vehicle occupant classification systems. Systematics, hardware environment and validation tools
Absicherung der Funktionalität optischer Sicherheitssysteme im Fahrzeug-Innenraum zur Insassenklassifizierung
2007
14 Seiten, 6 Bilder, 2 Tabellen, 8 Quellen
Conference paper
English
Automotive engineering | 2007
|British Library Conference Proceedings | 2007
|