| With the development of computer and information automation technology,mobile robot is finding wider and wider application in all fields.With the development of the national economy and population decrease in rural areas,the agricultural labor intensity has become more intensive so that many scholars focus on agricultural mobile robot to reduce labor intensity.At the same time,hills and mountainous area in our country is large and the design of a mobile platform for hilly areas is of great practical significance.The purpose of the paper is to design and develop a mobile platform for hilly mountain.To run on the ramp with picking device in hilly areas,the size parameters of the platform have been improved based on Komodo-2 crawler chassis according to the specific requirements of Citrus harvesting robot picking state.The vision navigation technology was studied to meet the technical requirements of automatic navigation in citrus picking when the mobile platform was regarded as a walk device.The major research of this paper were summarized:(1)According to the operation characteristics of picking robot,the key size parameters to keep the platform stable are calculated when the manipulator moved in its maximum velocity were calculated after analyzing the motion inertia of manipulator joint,in which the hilly areas were taken into account.The crawler chassis of Komodo-2 was redesigned so that the calculation of the size parameters for whole platform and the layout of the picking robot were done.(2)According to the orchard road characteristics,the gray images in H channel were selected and the fixed threshold for image segmentation was determined in the light of the characteristics of orchard road image.The advantages of proposed segmentation method has been proved after comparing with three kinds of segmentation methods.this paper presents a method for distinguishing central road line to deal with the problem of how to generate a navigation line,in which linear fitting and three polynomial fitting were applied to straight road and curve roads respectively.Statistics of image recognition results suggest the success rate of path recognition of proposed method was 90.0%.(3)In order to make the navigation line show the real road information,a monocular camera positioning model was deduced from the camera imaging principle and coordinate transformation principle,which was able to acquire the target position information under certain conditions.The experiments showed the average error of the model was 8.5mm and 25.7mm in horizontal and vertical direction respectively.(4)In light of the characteristics of hilly terrain,in order to achieve stable and accurate point-to-point movement,a path planning method named arc-line-arc was proposed for the platform.the mobile platform was able to reach any target position with the path planned by the proposed method when the coordinate value on y axis was 3 times larger than the turning radius(l?3r).The correctness of theoretical derivation of the method was verified by simulation analysis and experiment.(5)The visual navigation system of the platform was designed using VB/MATLAB mixed programming method.When the mobile platform run at the speed of 0.72 m/s,the experiments showed the the maximum navigation offset for straight road and small curvature road was 0.1m and the the maximum navigation offset for sharp turn was 0.3m. |
PDF Full Text Request