Evaluation of Feature Selection for Anomaly Detection in Automotive E/E Architectures
As we move towards higher levels of automation in autonomous driving, we see an increase in functionality that either assists or takes over in both normal and emergency scenarios. These new functionalities can be switched off by the user for personalisation. We aim to recognise mistimed and/or unintended deactivation of vehicle functions, in particular, driver assistance functions (ADAS), at run-time. This will be done in addition to already applied methods at design time. Upon recognition of the occurrence, we propose to inform the user and the OEM in order to improve both the future and the current system behaviour, to support development processes. Based on eight customer datasets, we evaluated our approach on a total of 17 state-of-the-art ADAS functions per participant, yielding to a total of 136 runs. We observed that during 24 among them, the user de-activated the functions at least once for more than a few seconds. For 13 of these 24 runs, we were able to detect and flag possible non-nominal behaviour over the full trace.