Moving Traffic Light Detection via Motion-Pattern Frame Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Camera-based vision systems in automated vehicles face difficulties in detecting traffic signal states when the traffic signal is moving due to factors like wind, leading to challenges in determining the correct light state.
Innovation Solution
A traffic-light-detection system that combines a camera and radar to determine the motion pattern of a traffic light and selects a preferred image from a series of images based on this pattern, ensuring the light source is most directed at the camera, thereby accurately determining the light state.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a camera-based vision system is used to detect traffic signal state, then the system can visually determine the light state, but the detection reliability deteriorates when the traffic signal is moving due to wind
Solution Approach 1:
The system performs preliminary action by capturing a series of images over time before making the final detection decision. The controller stores multiple images and identifies the preferred image showing the light source at the mid-point of motion, where the light state is most clearly directed at the camera. This preliminary capture of multiple states allows the system to overcome the harmful effect of wind-induced motion and select the optimal detection moment.
2Measurement precision
If images are captured continuously to account for motion, then the detection accuracy improves, but the processing time and complexity increase
Solution Approach 1:
The controller performs preliminary analysis by examining a series of captured images to identify which one shows the light source at the mid-point of motion. This preferred image provides the most accurate representation of the light state. By pre-processing the image series to select the optimal frame, the system achieves high measurement precision while managing processing time efficiently through targeted analysis rather than processing all possible variations.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The system enables reliable and quick determination of the traffic light state even when the light is moving, improving the confidence level of detection and reducing errors caused by varying light intensity and position.
Implementation Method 1
a radar (28) on the host vehicle (12), said radar (28) configured to detect radar returns (30) reflected by the traffic light (20)
Data Source
Figure 1
Figure 2
Figure 3A~3B
AI summary
A traffic-light-detection system (10) that visually determines a light-state (18) of a traffic-light (20) proximate to an automated vehicle includes a camera (24), a controller (32), and optionally a radar (28). The camera (24) and the radar (28) are on a host- vehicle (12). The camera (24) renders a series-of-images (26) of a traffic-light (20) proximate to a host-vehicle (12). The radar (28) detects radar-returns (30) from the traffic-light (20). The controller (32) is configured to determine a motion-pattern (34) of the traffic-light (20) based on the series-of-images (26) and/or the radar-returns (30), and select a preferred-image (36) from the series-of-images (26) based on the motion-pattern (34). The preferred-image (36) shows a light-source (38) of the traffic-light (20) characterized as being most directed at the camera (24) when the motion-pattern (34) indicates that the traffic-light (20) is moving. The controller (32) is further configured to determine a light-state (18) of the traffic-light (20) based on the preferred-image (36).