Communications of the ACM
AI Helps Vehicles Detect Potholes in Real-Time
Images of eight road-surface types were used to train and test the model.
Credit: Getty Images
The Korea Institute of Civil Engineering and Building Technology has announced the development of an AI-based system to automatically detect potholes, which can damage cars or lead to accidents.
A research team at led by Seungki Ryu developed the system which detects potholes in real-time by photographing the road surface while driving with a vision sensor installed on a vehicle windshield. The AI inference model semantically segments damages on the road surface using an encoder-decoder network based on the fully convolutional neural network architecture.
A mobile app gathers data using the AI model and a map-based cloud server platform to identify potholes. "It is essential to maintain road facilities in good condition in the coming era of autonomous vehicles," Ryu says. Several local governments in Korea are piloting the technology.
The researchers describe the system in "CNN-Based Road-Surface Crack Detection Model That Responds to Brightness Changes," published in the journal Electronics.
From National Research Council of Science & Technology
View Full Article
No entries found