Pollutants Positioning And Clean Planning In The Vision Of Mobile Robot

Due to poor working conditions, the robot's visual system is very vulnerable to pollution,affecting the normal operation. Thus, robot vision systems need solutions for cleaning. This paper designed visual feedback system for automatic cleanup of mobile robot vision.Firstly, according to mission requirements, a mobile robot platform based on embedded systems is built, with a personal computer, which provides support for follow-up experiments.The work of hardware part to be done includes selections of cameras, carriers for embedded systems and controller; the work of software part to be done includes the establishment of embedded systems, video streaming and wireless communication capabilities.Secondly, in a complex environment, an algorithm is proposed for automatic delayed extraction of pollutants in the field of mobile robot. It includes the separation of pollutants from backgrounds with accumulated frame difference, determining whether the presence of contaminants by analyzing trends in the histogram of cumulative frame difference image,adaptive thresholding targets pollution with Otsu algorithm. This achieves automatic extraction of pollutants. Mathematical morphology operations are done on areas of target pollutants to remove noise and fill holes. Analyzing connectivity,contaminated areas are turned into Simply connected regions by extracting outer contours of contaminants. Convex hulls, smallest minimum bounding rectangles and minimum bounding rectangles of simply connected regions are obtained as reference areas to be cleaned. The areas and centroids of them are calculated by silhouette moment.Thirdly, analyzing the problem of clean path optimization and regarding the total length of the route to clean as a standard,cleaning efficiency is compared between various external polygons. The total length of the route to clean can be divided into the actual length of the cleaning route and the length of empty stroke. The actual length of the cleaning route is obtained by calculating the areas of the external polygons. The length of empty stroke optimization belongs to traveling salesman problem. Cleaning paths are designed for different external polygons and Particle swarm optimization based on nearest neighbor is used to optimize the length of empty stroke. Zhang's calibration method is used to get the internal parameters of the camera, and a system of equations for three-dimensional coordinates of pollutants is established through coordinate transformation. The system is solved by calling Matlab calculation engine in the Visual C + + environment for positioning pollutants.Three-dimensional models of lens hood and contaminants are built, and various external polygons back project for analysis and comparison.Finally, the impact of three forms of exercise for cameras, pan, horizontal translation and dollying on the algorithm for automatic extraction of pollutants is studied through experiments. The algorithm is made appropriate improvements. The clean route optimization is combined with automatic extraction of pollutants in mobile robot, to verify the effectiveness of the proposed method.