Vehicle Localization Using Section-Based Image Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current motor vehicle localization methods in environments are not sufficiently precise, especially in complex scenarios, leading to inefficiencies and potential safety issues during semi-automated or automated driving.

Innovation Solution

A system comprising a camera-equipped device in the vehicle that captures image sequences, recognizes and classifies objects, defines sections, and transmits lane information using polynomial functions to a map management device for precise localization, which compares and updates digital maps, enabling efficient and accurate vehicle positioning.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If detailed map data and multiple sensors are used for localization, then localization precision is improved, but data processing complexity and computational load increase

Engineering Contradiction:
Improvelocalization precisionVSAvoiddata processing complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the environment into discrete sections with predetermined boundaries and sizes. Each section contains specific objects that are recognized and classified independently. This segmentation allows the system to process localization data in manageable units rather than handling the entire environment as one complex dataset, thereby reducing computational complexity while maintaining precision.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts only the necessary features and objects relevant to localization from the captured image sequences. By identifying and extracting specific objects within defined sections rather than processing all visual data, the system reduces data processing complexity while preserving the essential information needed for accurate vehicle localization.

Inventive Principle:
Principle #2Taking out (Extraction)

2Measurement precision

If comprehensive environment detection is performed, then localization accuracy is improved, but processing time increases

Engineering Contradiction:
Improvelocalization accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary actions by pre-defining sections with predetermined boundaries and sizes before actual localization occurs. Objects and their positions are pre-organized within these sections. When localization is needed, the system only needs to match current sensor data against these pre-processed sections, significantly reducing processing time while maintaining comprehensive environment detection for accuracy.

Inventive Principle:
Principle #10Preliminary action

3Quantity of substance

If polynomial functions are used to describe lane boundaries, then data requirements are reduced, but mathematical complexity increases

Engineering Contradiction:
Improvedata requirementsVSAvoidmathematical complexity
Core Design Contradiction:
Quantity of substanceVSDevice complexity

Solution Approach 1:

The patent changes the parameter representation of lane boundaries from detailed geometric descriptions to polynomial function parameters. Instead of storing numerous points defining lane boundaries, the system represents them using polynomial equations with fewer parameters. This parameter transformation reduces data requirements while the mathematical complexity is managed through the systematic application of polynomial fitting and evaluation methods.

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP3380810B1Method, device, map processing device and system for precise localization of a vehicle in its surroundings
Publication Date: 2022.04.06 VOLKSWAGEN AG
  • EP3380810B1 patent drawingFigure 1
  • EP3380810B1 patent drawingFigure 2
  • EP3380810B1 patent drawingFigure 3

AI summary

The invention relates to a method for precision-locating a motor vehicle (50) in an environment (12), comprising the following steps in a device (2) in the motor vehicle (50): capturing a sequence of images of the environment (12) of the motor vehicle (50) by at least one camera (4), detecting and classifying objects (28) in the captured sequence of images by an evaluating unit (5), determining object positions of the objects (28) relative to the motor vehicle (50) by the evaluating unit (5), defining a section (13) in the environment (12), wherein the section (13) has a predefined size and predefined boundaries (14-17), allocating the objects (28) to the determined object positions in the specified section (13), identifying a lane (22) of the motor vehicle (50) in the section (13), transmitting object information and the positions of the objects (28), section information on the defined section (13), lane information and time information to a map management apparatus (3) by a transmitting apparatus (6). The data is received and compared to a digital map (60) in the map management apparatus (3). Corresponding environmental maps are then transmitted to the device (2) and compared to environmental data of the defined section (13) and the vehicle (50) is located. The invention further relates to a device (2), to an associated map management apparatus (3), and to an associated system (1).