System for classification of weather conditions and detection of sensor failures (ASGARD-WECL)

The ASGARD-WECLASS software system is able to determine the weather conditions in which the sensor is currently working (fog, rain, reduced visibility, etc.) and detect some sensor failures (e.g. light glare, lens freezing, dirtiness, etc.). The evaluation takes place both from the sensors separately and from a combination of several sensors of the same type (overlapping area analysis) and from a combination of sensors of different types (data fusion). The output of the system is metadata, describing how reliable the data output from the sensor is. In addition to the above, it is able to detect failures in the area of the geometry of the sensors, for example in the event of a vehicle impact. The main use of the system is in autonomous and semi-autonomous vehicles, for which robust and reliable behavior is important. Information about the degree of reliability of the data that the system receives from the sensors is essential for such robust behavior.