Vehicle Localization Using Static Object Pattern Filtering

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for localizing and mapping vehicles in environments, such as SLAM algorithms, face challenges in maintaining accuracy due to slight environmental changes like snowfall, growing grass, or falling leaves, which can result in localization errors.

Innovation Solution

A method that uses environment perception sensors to identify and filter measurements based on predefined patterns of static objects offset from each other, allowing for more robust and reliable localization and mapping by disregarding minor environmental changes.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If SLAM algorithms are used for localization and mapping, then the vehicle can be localized in the environment, but localization accuracy deteriorates due to slight environmental changes like snowfall, growing grass, or falling leaves

Engineering Contradiction:
Improvelocalization accuracyVSAvoidenvironmental changes
Core Design Contradiction:
ReliabilityVSObject-affected harmful factors

Solution Approach 1:

The patent segments the environment into predefined patterns of static objects (such as poles, signs, or other structured elements) that are offset from each other in a specific configuration. By focusing only on these segmented, pattern-matched objects rather than the entire environment, the system filters out irrelevant changes like snowfall or falling leaves, thereby maintaining localization accuracy despite environmental variations.

Inventive Principle:
Principle #1Segmentation

2Measurement precision

If environment perception sensors continuously obtain measurements for localization and mapping, then the vehicle position can be obtained, but processing power requirements increase

Engineering Contradiction:
Improvevehicle position accuracyVSAvoidprocessing power
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent extracts and filters only those measurements that correspond to predefined patterns of static objects with specific offset relationships. By taking out only the relevant pattern-matched objects from the continuous sensor measurements and discarding other data, the system reduces the processing load while maintaining measurement precision for vehicle localization.

Inventive Principle:
Principle #2Taking out (Extraction)

3Quantity of substance

If all objects in the environment are used for localization, then more data is available for mapping, but localization errors increase due to moving objects and environmental changes

Engineering Contradiction:
Improvenumber of objectsVSAvoidlocalization accuracy
Core Design Contradiction:
Quantity of substanceVSReliability

Solution Approach 1:

The patent applies local quality by assigning different properties to different objects based on their characteristics. Predefined patterns of static objects with specific offset relationships are identified as suitable for localization, while other objects (moving objects, transient elements) are excluded. This selective approach ensures that only reliable, stable objects contribute to localization, improving accuracy while maintaining a sufficient quantity of reference objects.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP4235337B1A method for localizing and/or mapping during operation of a vehicle in an environment
Publication Date: 2024.08.14 VOLVO AUTONOMOUS SOLUTIONS AB
  • EP4235337B1 patent drawingFigure 1~2
  • EP4235337B1 patent drawingFigure 3~4
  • EP4235337B1 patent drawingFigure 5a~5b

AI summary

The invention relates to a method for localizing and/or mapping during operation of a vehicle (100) in an environment, wherein the vehicle (100) comprises at least one environment perception sensor (110) for the localizing and/or mapping, the method comprising: - obtaining (S1) at least one measurement of the environment from the environment perception sensor (110), - searching for and identifying (S2) a plurality of objects (10, 20, 30) in the at least one measurement which correspond to a predefined pattern, - in response to determining that the plurality of objects correspond to the predefined pattern, filtering (S3) the at least one measurement so that only the plurality of objects which correspond to the predefined pattern are used for the localizing and/or mapping, and - localizing and/or mapping (S4) based on the filtered at least one measurement. The invention also relates to a control unit (120), a vehicle (100), a computer program and a computer readable medium.