Robust Traffic Light and Arrow Detection Using Digital Map with Spatial Prior Information for Automated Driving.

Area of Science:

• Computer Vision
• Robotics
• Artificial Intelligence

Background:

• Traffic light recognition is crucial for urban automated driving.
• Existing methods face challenges with detection accuracy and computational cost.

Purpose of the Study:

• To propose an algorithm for recognizing traffic lights and arrow lights using image processing.
• To enhance automated driving systems with reliable signal detection.

Main Methods:

• Utilizing digital maps and precise vehicle pose estimation to define regions-of-interest.
• Developing an algorithm for arrow light recognition based on relative traffic light positions.
• Employing image processing techniques for robust detection.

Main Results:

• Achieved an average f-value of 91.8% for traffic lights and 56.7% for arrow lights.
• Successfully detected small arrow objects (under 10 pixels).
• Performance meets requirements for smooth automated driving maneuvers at intersections.

Conclusions:

• The proposed algorithm effectively recognizes traffic and arrow lights.
• The method enhances automated driving capabilities by providing reliable visual cues.
• Digital map integration and relative positioning improve detection robustness and efficiency.