Lateral Positioning Method for Unmanned Roller Compactor Based on Visual Feature Extraction
Yiming Sun, Hui Jun Xie · 2019
Accurate and real-time lateral positioning is the key to path tracking control of unmanned roller compactor. In order to ensure the accuracy and real-time of lateral positioning, a lateral positioning method based on visual feature extraction is proposed. Firstly, the camera calibration was carried out to obtain its internal parameters and external parameters. Then the image processing method was used to detect and fit the ground markings in the working area. Finally, in order to achieve the lateral positioning, the visual positioning model was used to calculate the distance between the look-ahead point in the moving direction of the roller compactor and the left and right ground markings. The experimental results show that the average absolute error of static positioning is 0.04 meters, the average relative error is 2.31%, the average absolute error of dynamic experiment is 0.06 meters, the average relative error is 3.51%, and the average processing speed of the algorithm reaches 70 frames/second. It fully meets the needs of positioning measurement accuracy and real-time in engineering applications. In addition, the method is less subject to the interference of electromagnetic signal and requires less hardware cost.