Traffic Signal Recognition Using Map-Based Light Position Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for detecting traffic signals struggle to accurately distinguish them from other lighting devices, especially at night when the housing of the traffic signal is not visible, leading to difficulties in determining the correct traffic signal among multiple light sources.
Innovation Solution
A traffic signal recognition device and method that utilize a vehicle position acquisition unit, an on-map position acquisition unit, and an observed position acquisition unit to determine the proximity between the detected lighting device and the registered traffic signal position in map data, enhancing detection accuracy by comparing the proximity of each lighting device.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If traffic signal detection is performed by estimating position using map data, then detection coverage is improved, but detection precision deteriorates due to inability to distinguish from other lighting devices
Solution Approach 1:
The patent introduces map data as an intermediary to bridge the gap between observed lighting positions and actual traffic signal positions. By comparing observed lighting positions with pre-stored traffic signal positions from map data, the system can accurately identify traffic signals even when their housings are not visible, thus improving detection precision while maintaining wide detection coverage
Solution Approach 2:
The system uses feedback by comparing the observed lighting position with the estimated traffic signal position from map data. When the proximity between observed lighting position and map-based traffic signal position matches within a predetermined range, the system confirms the lighting device is a traffic signal, thereby improving detection accuracy without limiting detection coverage
2Reliability
If multiple lighting devices are detected, then detection sensitivity is improved, but identification accuracy deteriorates due to difficulty in determining the correct traffic signal
Solution Approach 1:
The patent applies local quality by evaluating each detected lighting device individually based on its proximity to the estimated traffic signal position from map data. Instead of treating all lighting devices uniformly, the system selectively identifies traffic signals by comparing each lighting device's position with the map-based estimate, thereby maintaining high identification accuracy even when multiple lighting devices are present
Solution Approach 2:
Map data serves as an intermediary that provides the estimated traffic signal position, enabling the system to distinguish the correct traffic signal from other lighting devices. By using this intermediary reference, the system can accurately identify traffic signals even in scenarios with multiple lighting sources, resolving the contradiction between detection sensitivity and identification accuracy
Data Source
AI summary
A traffic signal recognition device recognizes a traffic signal present around a vehicle. The device acquires a position of the vehicle, and acquires a position of the traffic signal registered in map data based on the acquired position of the vehicle. The device acquires a position of a lighting device detected using an image frame captured by an in-vehicle camera. The device determines whether the lighting device corresponds to the traffic signal based on a difference between the acquired position of the traffic signal and the acquired position of the lighting device.