Method and Device for Processing Sensor Data in a Vehicle

autonomous-driving
A modeling unit for autonomous vehicles that fuses data from multiple heterogeneous sensors into an environment model and propagates per-sensor ASIL integrity information through to the model, enabling vehicle functions to dynamically adapt their automation level to the current integrity of the environment model.
Authors

Luca Parolini

Sebastian Schneider

Published year

2019

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

Overview

The present disclosure describes a modeling unit for constructing an environment model of a vehicle’s surroundings. The modeling unit receives sensor data from a plurality of heterogeneous sensors — including radar, lidar, camera, and ultrasonic sensors as well as vehicle state sensors — together with integrity information (ASIL level) associated with each sensor’s data. From these inputs, the modeling unit derives both the environment model and integrity information characterizing the reliability of that model.

The environment model and its associated integrity information are updated continuously at each time step. A downstream function unit uses the integrity information to dynamically adapt the degree of automation of vehicle functions: as the integrity of the environment model decreases — for example due to sensor degradation or the failure of a processing module — the automation level is reduced accordingly, enabling safe and flexible operation across SAE levels 1 through 4.

Filings

Jurisdiction Kind Application / Publication no. Date
DE Application DE102019134766.1A 2019-12-17
DE Publication (A1) DE102019134766A1 2021-06-17

Related