Research On Visual Navigation And Path Tracking Of Mobile Robot Based On Marking Line

Navigation technology is one of the key technologies in mobile robot research.Visual navigation is one of the main development directions of mobile robot navigation because of its abundant information,high sensitivity and good scalability.Compared with other methods such as three-dimensional visual recognition technology,it is easy to set and change the route of the marking line with the low cost,and the image processing based on marking line is fast,real-time and reliable.In this paper,aimed at visual navigation and path tracking based on marking line,related research is implemented.An efficient method of marking line extraction is proposed for the problem of path extraction in image.Meanwhile,in order to improve the stability of robot operation,a stable filtering and efficient fuzzy control method is designed,and the feasibility is verified in different scenes.The detailed contents are as followed:Firstly,in view of the problem of noise and environmental interference in image,a method of extracting marking line from complex environment is proposed.The input image is processed by Gauss filter,the outline of the marking line is detected by Canny operator,and the edge line of the marking line is extracted by Hough transform,finally,the area of marking line path is extracted from the edge line by FloodFill algorithm.The result path information has high noise resistance.At last,the path area is refined by morphological filtering method,and the center line of the marking line is extracted.Experiments show that the method can extract the center line of the marking line with high accuracy.Secondly,in order to improve the reliability of image processing of marking line,digital signposts are set up in places where need to stop or turn,and the KNN classifier is used to train the sample of digital signposts,so that the robot can change its tracking strategy after recognizing the digital signposts during the running process,which reduces the processing time of image to some extent.When facing multiple crossing paths,the path information is coded beforehand,the robot can turn correctly according to the coded path information when facing multiple crossing paths.Then,preview point is used to track the marking line in the image space,according to the result of image processing and recognition,the robot model and motion model are established,Kalman filter and fuzzy control are combined to realize the deviation correction control of the robot operation.Experiments show that control method can achieve the goal of tracking the marking line,and the operation is stable.Finally,the development platform based on ROS is built,image processing and recognition module and path tracking module are designed and implemented.The feasibility of the design method in this paper is verified in straight path,turning path and obstacle scenes.