… unstudied problem for self-driving cars. This paper introduces a novel automated method
to identify mistakes made by object detectors on the raw unlabeled perception data streams …

… We study two distinct cues to identify errors - temporal and … Temporal: CNN based object
detectors often fail to … baseline setup, object detectors often fail to consistently detect objects in …

… Selfdriving cars making mistakes will be an inevitable part of … failure prediction work mostly
focuses on detecting failures … of scene complexity as perceived by the car, we also use the the …

… learning to detect patterns that indicate impending failures. … Different from those works, we
predict failures of an entire self-driving … [14] detect failures of a perception module by running a …

… failures to date: the 2018 fatal crash of Uber's self-driving car. … constant comparative analysis
to identify key sources and … SDS meant that repeated failures of the perception system to …

… The perception part is typically delegated to deep neural … failures, based on the Udacity
simulator for self-driving cars. This dataset can be used to evaluate the performance of failure …

… we used the Udacity simulator for self-driving cars [24], a cross-… exhibits multiple failures,
we assess each failure individually. … Yiqiang, “Monitoring perception reliability in autonomous …

… to identify outliers that are far away in the feature space. … Self-Oracle is unable to distinguish
between images it has … We find that VAEs often obtain similar reconstruction errors on OOD …

… based perception systems are susceptible to errors, eg, failing … solutions for the introspection
of perception errors in ADSs. We … , with a focus on self-driving vehicles, shared mobility, and …

… task in environment perception for autonomous driving, but … provide specific object-wise
failure predictions, which we believe … driving, we prioritize FNs among other types of failures, and …