Use of Bayesian Network for Detecting Sensor Blockage

autonomous-driving
A method for constructing a Bayesian network that detects blockage of multiple vehicle sensors by combining prior probabilities of sensor blockage and target presence with conditional detection probabilities.
Authors

Sebastian Schneider

Luca Parolini

Published year

2020

Inventors
Sebastian Schneider, Luca Parolini
Patent no.
Assignee
Bayerische Motoren Werke AG

Abstract

A method for generating a Bayesian network configured to detect blockage of a plurality of sensors of a vehicle. The method includes setting a set of a priori probabilities comprising: an a priori probability that a target exists in the field of view of the sensors; for each sensor, an a priori probability that the sensor is blocked; and for each sensor, an a priori probability that the sensor is not blocked. The method further sets conditional probabilities covering all combinations of blocked/unblocked sensor and presence/absence of the target in the sensor’s field of view, and generates the Bayesian network from these priors and conditionals.

Filings

Jurisdiction Kind Application / Publication no. Date
EP Application EP20191667.3 2020-08-19
EP Publication (A1) EP3958186A1 2022-02-23

Related