Traffic Signal Detection Region Correction via Lamp Candidates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing traffic signal detection systems fail to accurately detect signals due to errors in the current position of the vehicle, leading to potential misalignment of the detection region and missed detections.
Innovation Solution
A traffic signal detection device that sets a third detection region by correcting the first detection region based on the relative position of traffic signals from lamp candidates detected in the actual environment, using landmark and map information, and camera data to improve detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If the detection region is set based on predicted traffic signal position and vehicle current position, then processing load is reduced and target region is limited, but detection accuracy deteriorates when vehicle position error occurs
Solution Approach 1:
The system performs preliminary detection of lamp candidates in the entire image before setting the final detection region. This preliminary action identifies potential traffic signal locations, which are then used to correct the predicted position and establish an accurate detection region, ensuring the traffic signal is captured even when vehicle position error occurs
Solution Approach 2:
The system uses detected lamp candidates as feedback to correct the predicted traffic signal position. The detected lamp positions provide feedback information that adjusts the detection region setting, creating a closed-loop system that compensates for vehicle position errors and maintains detection accuracy
2Productivity
If the image processing region is limited to reduce processing load, then productivity is improved, but reliability deteriorates when traffic signal deviates from the region
Solution Approach 1:
The system performs preliminary scanning of the entire image to detect lamp candidates before restricting processing to a specific detection region. This preliminary action ensures no traffic signal is missed, and the detected lamp positions are then used to define a refined detection region for final processing
Solution Approach 2:
The detection region is dynamically adjusted based on detected lamp candidates. Instead of using a fixed predicted region, the system adapts the detection region location and size according to actual lamp detections, making the processing region flexible and reliable under varying conditions
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A traffic signal detection device 100 includes: a traffic signal position estimation unit 13 which estimates a relative position of a traffic signal with respect to a vehicle 5, based on map information D02 including positional information of the traffic signal around the vehicle 5 and a self-position of the vehicle 5; a first detection region setting unit 21 which sets a first detection region on an image for each traffic signal predicted to be present on the image obtained by an image capturing unit 11 from the relative position of the traffic signal; a second detection region setting unit 14 which sets, when it is predicted that a plurality of traffic signals is present on the image, a second detection region including the plurality of predicted traffic signals on the image; a third detection region setting unit 23 which sets a third detection region by correcting the first detection region, based on lamp candidates of the traffic signals detected from the second detection region; and a traffic signal detector 17 which detects the traffic signals from the third detection region.