Possibilistic Causal Reasoning Approach to Functional Deficiency Diagnosis of Automated Driving System

Abstract

Aiming at evaluating sufficiency of functional deficiency knowledge in automated driving system, this paper discusses a diagnosis problem during analysis of empirical database. A structured knowledge model is previously defined by a relation space between boundary (challenging operating conditions) and trigger-event (unwanted subsystem functionality). The proposed approach works on a set of observed trigger-events and performs a two-fold diagnosis task: (i) inference of boundary plausibility; (ii) classification of observation explainability. As illustrated, it could be a promising approach for large-scale case, which is being developed in our ongoing work.