Monitoring of perception systems: Deterministic, probabilistic, and learning-based fault detection and identification

Run-Time Introspection of 2D Object Detection in Automated Driving Systems Using Learning Representations

“…Inconsistency-based Introspection: Leveraging the multimodality of ADS perception, these methods declare faults when outputs from different algorithms or sensors don't match. Techniques vary from using stereo and temporal cues to graph-based methods for identifying inconsistencies between processors or data inputs [27]- [29].…”