Road Element Signature Clustering for Precise Scene Classification

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing assisted and autonomous driving systems face challenges in efficiently classifying road elements due to the need for improved classification systems and methods.

Innovation Solution

A method and system utilizing neural networks to generate dynamic in-correlation signatures through iterative processes, where identifiers are generated based on relative occurrences in true and false positive signature sets, allowing for accurate and robust classification of road elements by associating multiple signatures per element.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional classification methods are used for road elements, then the system complexity is low, but the classification accuracy and precision are insufficient

Engineering Contradiction:
Improveclassification accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the classification process into multiple iterative stages, generating multiple signatures (first signature, second signature, etc.) with different identifier sets. Each signature focuses on different aspects of road elements, and their combinations improve overall classification accuracy while managing system complexity through modular processing

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces a new dimension of classification by generating multiple signatures with different identifier combinations rather than relying on a single classification approach. This multi-signature approach adds dimensional depth to the classification process, enabling more precise differentiation of road elements

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

2Manufacturing precision

If multiple signatures are generated per road element to improve classification accuracy, then the precision of road object detection improves, but the computational complexity and processing time increase

Engineering Contradiction:
Improvedetection precisionVSAvoidprocessing time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary actions by pre-generating multiple signature sets with different identifier configurations before actual classification needs arise. These pre-computed signatures are stored and can be rapidly applied during runtime, reducing processing time while maintaining high detection precision

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent changes parameters by varying the identifier sets across different signatures (e.g., first identifiers, second identifiers with different properties). This parameter variation allows the system to explore multiple classification perspectives simultaneously, improving detection precision without linearly increasing processing complexity

Inventive Principle:
Principle #35Parameter changes

Data Source

PatentEP4668227A1Dynamic in-correlation signature generation
Publication Date: 2025.12.24 AUTOBRAINS TECH LTD
  • EP4668227A1 patent drawingFigure 1
  • EP4668227A1 patent drawingFigure 2
  • EP4668227A1 patent drawingFigure 3

AI summary

A method that includes (i) generating, in a first iterative process, a first signature comprising first identifiers that are indicative of at least one of (a) a feature of a road element associated with the first signature or (b) a feature of a generation of the first signature, the first identifiers being generated in correlation to each other, and (ii) generate, in a second iterative process, a second signature comprising second identifiers that are indicative of (a) a feature of a road element associated with the second signature or (b) a feature of a generation of the second signature, the second identifiers being generated in correlation to each other; wherein the second identifiers of the second signature are generated in de-correlation to the first identifiers of the first signature, and wherein the first signature and the second signature collectively represent a cluster of sensed information.