Descriptor-Based Vehicle Map Positioning in Semantic Road Maps

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing methods for positioning a vehicle's surroundings in a semantic road map lack precision and efficiency, particularly in updating and integrating detailed surroundings data.

Innovation Solution

A method involving the creation of a map representation using surrounding sensor data, comparison of characteristic elements with a semantic road map, and utilization of descriptors like neighbor binary landmark descriptors and RANSAC algorithm for precise alignment and integration of surroundings data into the semantic road map.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Device complexity

If characteristic elements are compared directly without descriptors, then the comparison process is simpler, but positioning precision deteriorates

Engineering Contradiction:
Improvecomparison process complexityVSAvoidpositioning precision
Core Design Contradiction:
Device complexityVSMeasurement precision

Solution Approach 1:

Descriptors serve as intermediary representations of characteristic elements. Instead of directly comparing raw characteristic elements between map representation and semantic road map, the system extracts descriptors (feature vectors capturing geometric and semantic properties) as intermediate forms. This intermediary step enables precise comparison while maintaining manageable complexity, as descriptors standardize the comparison format and highlight key distinguishing features.

Inventive Principle:
Principle #24Intermediary (Mediator)

Solution Approach 2:

The system transforms characteristic elements into descriptors by changing their representation parameters. Descriptors encode multiple parameters (position, orientation, shape, semantic class) of characteristic elements into standardized feature vectors. This parameter transformation enables quantitative comparison and precise matching between map representation and semantic road map elements.

Inventive Principle:
Principle #35Parameter changes

2Loss of information

If the map representation covers the entire surroundings, then completeness is improved, but processing time and computational load increase

Engineering Contradiction:
Improvesurroundings coverage completenessVSAvoidprocessing time
Core Design Contradiction:
Loss of informationVSLoss of time

Solution Approach 1:

The system segments the surroundings into multiple map representations, each covering a specific field of view or region. Instead of processing one large comprehensive map, the system divides the environment into manageable segments (e.g., based on sensor fields of view or geographic regions). Each segment is processed independently for positioning, reducing computational load per task while maintaining overall completeness through aggregation of multiple segments.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system performs partial positioning actions by comparing only relevant characteristic elements within each map representation against the semantic road map, rather than processing all possible elements. By focusing computation on salient characteristic elements (road markings, signs, buildings) within each segment, the system achieves sufficient positioning accuracy without exhaustive processing of all surroundings data.

Inventive Principle:
Principle #16Partial or excessive action

3Measurement precision

If more characteristic elements are used for positioning, then positioning accuracy is improved, but the complexity of identifying and comparing elements increases

Engineering Contradiction:
Improvepositioning accuracyVSAvoidelement identification and comparison complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system applies different processing qualities to different characteristic elements based on their importance and reliability for positioning. Not all characteristic elements are treated equally - the system identifies and prioritizes high-value elements (such as unique road signs, distinctive buildings, or well-defined road markings) that provide strong positioning signals. Less important elements are processed with simpler methods or given lower weight in the positioning calculation, optimizing the accuracy-complexity tradeoff.

Inventive Principle:
Principle #3Local quality

Data Source

PatentEP4399488B1Method for positioning a map display of surroundings of a vehicle in a semantic road map
Publication Date: 2025.10.15 ROBERT BOSCH GMBH
  • EP4399488B1 patent drawingFigure 1~2
  • EP4399488B1 patent drawingFigure 3~4
  • EP4399488B1 patent drawingFigure 5

AI summary

The invention relates to a method (100) for positioning a map display (200) of surroundings (501) of a vehicle (500) in a semantic road map (300), comprising: - receiving (101) surroundings sensor data (505) from at least one surroundings sensor (503) of a vehicle (500); - creating (103) a map display (200) of surroundings (501) of the vehicle (500) based on the surroundings sensor data (505) of the vehicle (500); - determining (105) characteristic elements (201) in the map display (200) and characteristic elements (301) in a semantic road map (300) depicting the surroundings (501); - comparing (107) the characteristic elements (201) of the map display (200) with the characteristic elements (301) of the semantic road map (300); - determining (109) a section (303) of the semantic road map (300) for which at least one characteristic element (301) of the section (303) of the semantic road map (300) matches at least one characteristic element (201) of the map display (200) and for which the map display (200) and the semantic road map (300) describe an identical region of the surroundings (501) of the vehicle (500); and - identifying (111) a position (P) of the section (303) within the semantic road map (300) as position of the map display (200) in the semantic road map (300).