Vision-based Obstacle Crossing And Line Grasping For Power Line Deicing Robot

The safe and stable operation of the high voltage power line is the basic guarantee of national economic development and people's life. However, in the snow and ice weather, there are often inverted tower and power outages and other accidents caused by high voltage overhead power line icing seriously. With the development of artificial intelligence and its application in various fields, it has a lot of positive effects on society. It is also an important direction for the development of artificial intelligence to introduce intelligent robots to the power line, and it has a positive significance.In this paper, firstly some research cases of robot application to power line operation in recent years are reviewed, and the current research results and some problems of power line deicing robot is introduced. For the research on visual control method of deicing robot autonomous obstacle in the power line, the structure and system of the three arm deicing robot of Special Robot Laboratory of Hunan University is introduced, and its kinematics control model is established. For the problem of robot positioning process on power line, two kinds of power line positioning method based on texture features are proposed. One is the introduction of Gabor filter to enhance and segment the power line texture region in the surveillance camera image, based on the advantages of Gabor filter in the texture enhancement and segmentation, so as to realize the positioning effect of the power line; another method is according to characteristics of LBP Operator on the sensitivity of texture features, the LBP map of the image is established, and the region with similar texture feature is extracted by the method of histogram statistics and fuzzy C means clustering, and the positioning effect of the power line is also achieved.Finally, for deicing robots in independent line grasping control problem, three kinds of servo control method based on image features are proposed in this paper. Firstly, the practical problems of grab line motion of deicing robot is analyzed, the corresponding image features are selected, and the Jacobian matrix based on image features is deduced, then the application of traditional servo control method in robot arm grasping line problem of deicing robot is established; and then the traditional servo control method is improved by introduced the wavelet neural network instead of the Jacobian matrix, the visual servo control system model is simplified, and the effectiveness of the method is proved by the line grasp experiment of deicing robot; in view of the defects of large sample collection based on the wavelet network servo control method, finally, a visual servo control method based on SVR witch using SVR instead of wavelet neural network is proposed in this paper. this method is more practical in the robot servo control system because of SVR has the advantages of small sample training and strong generalization ability, and the effectiveness of the method is proved by the experiment of the robot's grasping line control.