Research On Robot Visual Navigation Of Agricultural Greenhouse

With the development of modern urban agriculture and intelligent manufacturing technology,agricultural robots become a typical representative application of intelligent control technology in agricultural equipment.The research of vision navigation system has been paid more and more attention by scholars.Comparing to farmland,greenhouse is easier to realize precise agriculture.Using mobile robot instead of human to work in the high temperature,high humidity toxic environment automatically and precisely,which can reduce environment pollution and the risk of poisoning of workers.At the same time,the utilization of pesticides can be improved effectively.So,intelligent greenhouse mobile robot has become one of the research focus both domestic and foreign scholars.In recent years,with the rapid development of Image processing technology and computer processing capacity,visual navigation has became one of the main direction of robot navigation.The purpose of this paper is realizing the autonomous navigation of greenhouse robot and the algorithm of navigation path of mobile robot is studied deeply.According to the characteristics of greenhouse crops planted in rows,the navigation path recognition algorithm is proposed by using computer image processing.First,this paper summarizes the research content,development status of vision navigation and the application of agricultural robot in greenhouse.Then the background of choosing subject and the primary structure of the paper wereSecond,the paper gives the summarize of the ways,classification and development of the visual navigation.The technology of visual navigation is described in detail,including the composition,the key technology and the working process of vision system.Third,the paper expounds image processing technology,which include image acquisition,pre-processing and image segmentation.To wipe off yawp,it puts forward the arithmetic which uses the improved median filtering and wavelet denoising method to eliminate mixing yawp.The result proveds that it can gain a full image information and good real-time characteristic.Fourth,adopting a new arithmetic of image segmentation to separated the plants from the background.After comparing recombination of the operator in RGB color space and analyzing the mean and standard deviation,a method was confirmed: firstly,segmenting the G-B and R-B difference image and H-histogram by using Otsu segmentation method;and then merging three of segmentations together as a whole and filters the noise by median function.The algorithm was simple and robust,met real-time requirements,as well as easy to adapt to the complicated greenhouse environment.The impact of illumination could be reduced and the navigation targets could be identify accurately.Fifth,the paper puts forward the ameliorating arithmetic of Hough transform.It will improve the compute speed and reduce the storage space and the processing time is two-fifths of the improvement.The accurate localization of the robot could be realized by using the information taken from distribution plant and the posture of the robot.The visual navigation algorithm could be implemented in software platform.