Path Planning Method Of Orchard Sprayer Based On RTK-BDS And RGB-D Depth Camera

China is a big country in fruit industry.With the continuous expansion of the scale of fruit tree planting,the application of agricultural intelligent equipment in orchards has effectively reduced the labor intensity of operators and improved the working efficiency and accuracy of orchard operations.The orchard sprayer is a mobile platform that can move autonomously among the rows of orchards and spray pesticides.It combines different end effectors to accomplish specified tasks in the orchard.Due to the complex and changeable environment in the orchard scene,the use of a single sensor to obtain environmental information is limited and cannot meet the navigation accuracy requirements.Therefore,in this paper,multi-sensors work together to obtain rich orchard information to achieve reliable and stable orchard navigation operations.In this paper,the Beidou navigation system and RGB-D depth camera are used as navigation perception devices to realize the autonomous navigation operation of orchard sprayer.The main research contents are as follows:(1)An autonomous navigation test platform for orchard spray robots was built.Based on the crawler robot,the perception control system of the orchard spray robot is designed.The orchard environment information is sensed through RTK-BDS and RGBD depth cameras to obtain the global map of the orchard and the position of the fruit trees,which provides the global and local path planning of the orchard spray robot data support.(2)Aiming at the problems of long sampling optimization time,random sampling points and new node generation of the basic RRT* algorithm,an improved Informed RRT* global path planning algorithm is proposed based on the orchard environment map constructed by RTK-BDS.On the basis of the Informed RRT* algorithm,the angle offset constraint and the reverse optimization strategy are introduced.The experimental simulation and comparative analysis are carried out in different complex environments.The overall environment information of the orchard is obtained through RTK-BDS to construct the orchard map for global route plan.The simulation results show that compared with the RRT* algorithm and the Informed RRT* algorithm,the improved Informed RRT* algorithm shortens the average path search time by 69.1% and 38.2%,respectively,and shortens the average path length by 6.1% and 2.4%,respectively.(3)In view of the problem that the BDS signal is lost due to the occlusion of the tree canopy during the driving process of the orchard spraying robot,which affects the navigation of the spraying robot between the rows of the orchard.A local path planning method for the orchard spraying robot based on the rapid identification of multi-feature tree trunks is proposed.Firstly,the color image and depth information of the orchard are obtained through the RGB-D depth camera.Then the preprocessed color image is divided into superpixels,and the adjacent superpixel blocks with similar color information and depth information are merged.Width feature,color features and parallel edge features are used for tree trunk recognition,and the accuracy of tree trunk recognition algorithm is verified under different lighting conditions.The tree trunk is located through the depth information,and the position of the fruit tree is identified,so as to generate the path fitting midpoint,fit into a road centerline,and then match the planned path.The global paths are combined to finally complete the path planning and navigation of the orchard spray robot.