Method for landmark-based localisation of a vehicle

Patent metadata

Field	Value
Publication number	US-11747159-B2
Application number	US-202117915641-A
Country	US
Kind code	B2
Filing date	Feb 19, 2021
Priority date	Mar 31, 2020
Publication date	Sep 5, 2023
Grant date	Sep 5, 2023

Abstract

A method for landmark-based localization of a vehicle involves forming a plurality of position hypotheses for a vehicle position based on a forming of associations between sensor landmark objects detected by sensor and map landmark objects stored in a digital map. A most likely vehicle position is ascertained as the localization result on the basis of a probabilistic filtering of the position hypotheses, and a guaranteed position area is ascertained, in which a predefined error upper limit is not exceeded. This is performed several times with different set-ups of the probabilistic filtering. The localization result with the smallest guaranteed position area is selected as vehicle position if the guaranteed position areas overlap fully in pairs.

First claim

The invention claimed is:

1. A method for landmark-based localization of a vehicle, the method comprising:
receiving, by an autonomous driving system of the vehicle, a determined current vehicle position; and
controlling the vehicle by the autonomous driving system so long as the determined current vehicle position has an integrity that is below an alert limit threshold value for the autonomous driving system of the vehicle, wherein the current vehicle position is determined by
a) forming, by the vehicle, a plurality of position hypotheses for a vehicle position based on associations formed between sensor landmark objects detected by a sensor of the vehicle and map landmark objects stored in a digital map;
b) probabilistically filtering, by the vehicle, the plurality of position hypotheses to produce a probability distribution of the vehicle position in the digital map;
c) determining, by the vehicle, a most likely vehicle position as a localization result based on probability distribution of the vehicle position in the digital map, and determining a guaranteed position area, in which a predefined error upper limit is not exceeded,
wherein steps a)-c) are performed several times with different configurations of the probabilistic filtering so that there are several localization results, wherein the different configurations of the probabilistic filtering relate to stringency of the configuration for discarding improbable position hypotheses, and
wherein one localization result of the several localization results from performing the steps a)-c) several times is selected as the determined current vehicle position if the guaranteed position areas from performing the steps a)-c) several times overlap fully in pairs, wherein the one localization result has a smallest guaranteed position area of the determined guaranteed position areas.

2. The method of claim 1 , wherein the different configurations of the probabilistic filtering relate to a feature association filter, in which the plurality of position hypotheses are filtered and the improbable position hypotheses are discarded.

3. The method of claim 1 , further comprising:
outputting an error message if at least one pair of the guaranteed position areas from performing the steps a)-c) several times overlap only in part.

4. The method of claim 1 , wherein the selected vehicle position (eFP) is assigned to all larger guaranteed position areas from performing the steps a)-c) several times.

5. The method of claim 1 , wherein the steps a)-c) are performed at least three times.

6. The method of claim 1 , wherein the performing of the steps a)-c) several times with different configurations of the probabilistic filtering is performed in a plurality of parallel branches of the method.

7. The method of claim 6 , wherein each of the plurality of parallel branches comprise a feature association filter, wherein the feature association filters of the plurality of parallel branches are configured differently, and wherein reliable position hypotheses is determined based on the feature association filters of the plurality of parallel branches.

8. The method of claim 7 , wherein, in each of the plurality of parallel branches, the reliable position hypothesis of the branch and an inherent movement of the vehicle are processed by a Kalman filter, and thus a corresponding localization result of each of the plurality of branches and a corresponding guaranteed position area of each of the plurality of branches, in which the predefined error upper limit is not exceeded, are determined.

9. The method of claim 1 , wherein sensor landmark objects already passed by the vehicle are used for the forming of the associations between the sensor landmark objects detected by sensor and the map landmark objects stored in the digital map.

10. The method of claim 1 , wherein the sensor landmark objects detected by the sensor over a predefined time period or over a predefined route length are used for the forming of the associations between the sensor landmark objects detected by sensor and the map landmark objects stored in the digital map.

Assignees

• Mercedes Benz Group Ag

Inventors

• Ries Florian
• Yang Yinglai
• Schuster Frank
• Haueis Martin

Classifications

• G01C21/3644 Primary

  Landmark guidance, e.g. using POIs or conspicuous other objects · CPC title

• G01C21/3874

  Structures specially adapted for data searching and retrieval · CPC title

• G01C21/30 Primary

  Map- or contour-matching · CPC title

• G01C21/1656

  with passive imaging devices, e.g. cameras · CPC title

• G01C21/3602

  Input other than that of destination using image analysis, e.g. detection of road signs, lanes, buildings, real preceding vehicles using a camera · CPC title