Traffic Sign Attribute Recognition for Unfamiliar Road Signs

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Autonomous vehicles face challenges in responding to traffic signs, especially unfamiliar ones, as existing systems rely on pre-mapped data which can be outdated or inaccurate, leading to potential safety issues and travel delays.

Innovation Solution

The method involves real-time detection of traffic signs using sensors and image processing techniques to determine attributes such as type, color, and content, allowing the vehicle to respond appropriately without prior identification in the map information, and prioritizing assistance requests based on sign type for safe operation.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If pre-mapped traffic sign data is used for identification, then the system can quickly recognize familiar signs, but it fails to identify unfamiliar or updated signs leading to safety issues

Engineering Contradiction:
Improvetraffic sign identification reliabilityVSAvoidability to recognize unfamiliar signs
Core Design Contradiction:
ReliabilityVSAdaptability or versatility

Solution Approach 1:

The system performs preliminary actions by capturing images of traffic signs with cameras before needing to identify them. This allows the system to have the sign image available for later analysis by the machine learning model, enabling identification of unfamiliar signs without requiring pre-mapped data.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

A machine learning model acts as an intermediary between the captured traffic sign image and the identification result. The model receives the image, analyzes it to determine sign type and attributes, and outputs the identification, bridging the gap between raw visual data and meaningful interpretation for unfamiliar signs.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If real-time image processing is performed to identify unfamiliar signs, then the vehicle can respond to unknown signs, but the processing time and computational resources increase

Engineering Contradiction:
Improvecapability to identify any traffic signVSAvoidtime for sign identification
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system captures traffic sign images in advance using cameras as the vehicle approaches, storing them for immediate processing when needed. This preliminary capture action eliminates delays that would occur from attempting to process or request images in real-time when identification is required.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system creates a digital copy of the traffic sign image through camera capture and stores it for processing. This copy can be immediately analyzed by the machine learning model without requiring physical interaction with the sign or real-time communication with external systems, significantly reducing identification time.

Inventive Principle:
Principle #26Copying

3Adaptability or versatility

If a machine learning model is used to determine sign type, then the system can identify unfamiliar signs, but it may produce false positives or negatives reducing accuracy

Engineering Contradiction:
Improvesign type identification capabilityVSAvoidsign identification accuracy
Core Design Contradiction:
Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The system implements feedback by comparing the machine learning model's identification results against the actual captured image and known traffic sign characteristics. When discrepancies or low confidence results occur, the system can request additional images, adjust processing parameters, or flag for manual review, continuously improving accuracy through feedback loops.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system performs preliminary verification by capturing multiple images or additional views of the traffic sign before final identification. This preliminary data collection allows the machine learning model to work with more comprehensive information, reducing false positives and negatives while maintaining adaptability to unfamiliar signs.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3877965B1Detecting unfamiliar traffic signs
Publication Date: 2024.10.09 WAYMO LLC
  • EP3877965B1 patent drawingFigure 1
  • EP3877965B1 patent drawingFigure 2
  • EP3877965B1 patent drawingFigure 3

AI summary

Aspects of the disclosure relate to determining a sign type of an unfamiliar sign 1053. The system may include one or more processors 120. The one or more processors may be configured to receive an image and identify image data corresponding to a traffic sign in the image. The image data corresponding to the traffic sign may be input in a sign type model 470. The processors may determine that the sign type model was unable to identify a type of the traffic sign and determine one or more attributes of the traffic sign. The one or more attributes of the traffic sign may be compared to known attributes of other traffic signs and based on this comparison, a sign type of the traffic sign may be determined. A vehicle 100 may be controlled in an autonomous driving mode based on the sign type of the traffic sign.