Traffic Light Grouping by Visual Relations at Complex Intersections
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in accurately recognizing and assigning traffic light structures over larger distances and at intersections with multiple signals, particularly when replicas are present, leading to errors and high computational demands.
Innovation Solution
A method that ascertains relations between activated traffic lights in video data, grouping them based on pixel size, position, and content-coded information to differentiate between original and replicated signals, reducing the need for 3D models and computational power.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If 3D models and complex computational methods are used to identify traffic light structures, then the recognition capability is improved, but the computational power required increases and recognition accuracy decreases at larger distances
Solution Approach 1:
The patent extracts and utilizes only the essential visual features (color, position, size) of traffic lights from video data, discarding the need for complex 3D modeling and extensive computational processing. This selective extraction of critical information reduces computational power requirements while maintaining recognition accuracy.
Solution Approach 2:
Instead of using complex 3D models to infer traffic light information, the patent inverts the approach by directly analyzing 2D video data characteristics (color, position, size) to identify and assign traffic lights. This inversion simplifies the computational process while improving effectiveness.
2Adaptability or versatility
If multiple traffic light structures including replicas are present at an intersection, then the system provides more comprehensive traffic control information, but the assignment of correct signals becomes more difficult and error-prone
Solution Approach 1:
The patent applies local quality analysis by examining specific characteristics (color, position, size) of each detected traffic light structure in the video data. By evaluating these local properties, the system can reliably distinguish between original traffic lights and replicas, and correctly assign signals to appropriate directions of travel even when multiple structures are present.
Solution Approach 2:
The patent utilizes color-coded information from traffic lights as a key distinguishing feature. By analyzing the color characteristics and their spatial relationships, the system can identify which traffic lights are originals versus replicas and correctly assign them to control specific directions of travel, resolving the ambiguity created by multiple structures.
3Loss of time
If traffic lights are recognized at larger distances, then the system provides earlier warning to drivers, but the recognition validity decreases and errors increase
Solution Approach 1:
The patent enables preliminary identification of traffic light structures by analyzing their characteristic visual features (color, position, size) in video data at larger distances. This preliminary action allows the system to identify and assign traffic lights earlier, providing advance warning to drivers while maintaining recognition validity through feature-based analysis rather than distance-dependent 3D modeling.
Data Source
AI summary
A method for grouping traffic light structures of a traffic light system, wherein each traffic light structure has at least one traffic light that can be activated, and wherein a first traffic light structure controls at least a first direction of travel. The method includes: ascertaining a relation of activated traffic lights to one another; grouping the traffic light structures taking into account the ascertained relation. A device and a corresponding computer program are also described.