This design is a pavement crack detection system, including pavement crack detection robot and human-computer interaction control software (PC or user receiver) two parts. The pavement crack detection robot with camera can automatically detect the pavement and mark and take photos of the suspected crack area, save the image and pavement position and call the relevant image processing algorithm to analyze the pavement crack image, including image enhancement algorithm, image denoising algorithm and image segmentation algorithm, and transmit the image information to the host computer (or user receiver) through the wireless transmission module. The upper computer system receives the images from the robot and uses the Alexnet network to classify the images, and calculates the width of the crack to obtain the damage degree of the crack, the location of the crack, the type and grade of the crack, and gives some repair suggestions. Technical workers can get specific information through the upper computer to replace the road maintenance workers to detect and mark the road environment. In addition, the system supports two modes of automatic exploration and detection and manual operation. The road crack detection robot itself adopts STM32F4 series main control chip, and the robot itself has a light supplement system, which can complete tasks in various environments such as sunny days, cloudy days and nights. The robot is powered by lithium battery module without connecting additional power supply wires.