LiCam: A Synergistic Approach for Pothole Detection

In this work we focus on developing a robust pothole detection system that integrates computer vision with LiDAR technology for reliable performance in varying lighting conditions. Potholes are a significant road hazard, contributing to vehicle damage, accidents, and fatalities worldwide. Traditional detection methods, such as camera-based systems, perform well in daylight but struggle in poor lighting conditions like night-time or fog. To address this limitation, we propose a cost-effective hybrid solution that combines camera for object detection with 2D LiDAR technology. Cameras provide high-resolution imagery and capture texture and shape information but are limited by ambient lighting. The addition of 2D LiDAR improves detection accuracy by measuring depth variations, ensuring reliable pothole detection even in low-light conditions. We proposed a system that captures multi-modal road surface data through camera and LiDAR, filtering out minor surface irregularities and focusing on significant dips. Testing under various lighting scenarios demonstrated that the system provides reliably accurate detections with real-time visual and audio alerts, improving road safety. While 3D LiDAR offers higher accuracy than 2D LiDAR, its high cost and computational demands make it less feasible for practical deployment. The 2D LiDAR approach balances accuracy with affordability, making it a viable alternative for large-scale pothole detection applications.