Onboard LiDAR-Camera Deployment Optimization for Pavement Marking Distress Fusion Detection

Area of Science:

• Road safety engineering
• Sensor fusion technology
• Artificial intelligence in transportation

Background:

• Pavement markings are essential for traffic guidance and safety but degrade over time.
• Periodic inspection of retroreflectivity and diffuse illumination is crucial for maintaining safety standards.
• Existing detection methods struggle with complex traffic conditions like shadows and changing light.

Purpose of the Study:

• To propose an optimized onboard sensor deployment method for pavement marking distress detection.
• To enhance the accuracy and robustness of pavement marking inspection systems.
• To adapt detection capabilities to dynamic environmental and traffic conditions.

Main Methods:

• Assessed the detection capabilities of Light Detection and Ranging (LiDAR) and camera sensors.
• Mathematically formulated the LiDAR-camera deployment optimization problem for distress fusion detection.
• Developed an improved Red Fox Optimization (RFO) algorithm with enhanced mechanisms to solve the deployment problem.

Main Results:

• Achieved a density of 5217 LiDAR points per data frame on pavement markings (0.58 m).
• Demonstrated a relative error of less than 7% between calculated and field test measurements.
• Validated the method's robustness in real-world scenarios, mitigating environmental interference.

Conclusions:

• The proposed sensor deployment optimization method effectively detects pavement marking distress.
• The system demonstrates high accuracy and robustness, even under challenging environmental conditions.
• This approach contributes to improved road safety through reliable pavement marking inspection.