Road Crack Acquisition and Analysis System Based on Mobile Robot and Deep Learning

In this paper, a road crack acquisition and analysis system based on mobile robot and deep learning is proposed. First, a virtual reality technology-based omnidirectional mobile robot is developed to remotely collect detailed road crack images. Next, an image dataset including different kinds of crack images collected by the mobile robot is constructed and utilized for training and testing a deep convolutional neural network (DCNN) model. Following this, the trained model is used to identify and segment the crack images acquired from the robot, and to facilitate the inspector to know the road surface condition, crack information (crack length, crack width, and crack area) is obtained using an image processing method. Finally, the proposed system and method are applied to a campus road, which realizes the accurate acquisition and analysis of road crack information, thus demonstrating the feasibility and effectiveness of the proposed system and method.