Research On Orchard Transportation Robot Navigation System Based On Multi-sensor

Orchard automatic navigation is an important part of the current orchard agricultural intelligent equipment.According to the characteristics of orchard planting,this paper develops an automatic navigation system based on the self-developed differential wheeled orchard transportation robot test platform.The research is carried out from the following three aspects:orchard environment perception positioning,navigation path planning and decision control.The specific contents are as follows:(1)The orchard transportation robot test platform used in this paper is designed and built.According to the actual orchard environment,the required control hardware is determined:GNSS sensor,2D laser sensor and industrial computer.The accuracy and stability of GNSS sensor and 2D laser sensor are analyzed.After experimental analysis,the average data of each frame of 2D laser sensor is at 1997 points,and the average distances to the left and right walls are 1422.948 mm and 1472.038 mm,respectively.The standard deviations are 0.985 mm and 0.949 mm,respectively,indicating that the 2D laser sensor is stable and meets the positioning accuracy.Through single-point static sampling of GNSS sensor,the maximum deviations of X-axis,Y-axis and heading angle are 0.075 m,0.046 m and 0.188°,respectively,indicating that the GNSS sensors used are accurate and stable.Finally completed the physical construction,to do the foundation for the following research.(2)Aiming at the complex environment of orchard,the orchard environment perception method was studied.Firstly,the fruit tree information acquisition method based on 2D-ICP is proposed.After iterative registration of point cloud data of both sides of fruit trees obtained by 2D laser sensor,the point cloud data of each fruit tree in the orchard are obtained.Then a fruit tree detection algorithm based on LAPO-DBSCAN is proposed to obtain the clustering center of each fruit tree and realize the perception of fruit trees.Through the analysis of simulation data,the positive detection rate of LAPO-DBSCAN algorithm is 2.42%higher than that of LAPO algorithm,and the false detection rate is 2.42%lower,and the time-consuming is reduced by 82.92%.Compared with the previous improved DBSCAN algorithm,the positive detection rate of the algorithm is increased by 3.92%,so the LAPODBSCAN algorithm is superior to the LAPO algorithm and the DBSCAN algorithm.Through the experimental analysis of the actual fruit tree data,the positive detection rate of LAPO-DBSCAN algorithm can reach 96.69%,the false detection rate is low to 3.31%,and the time is 1.14 s,indicating that the algorithm can be used for fruit tree detection to obtain fruit tree position.(3)Aiming at the problem of orchard positioning,the positioning method of transport robot is studied.GNSS sensor and 2D laser sensor are used for positioning.GNSS sensor is used for the positioning of the transport robot in the orchard.After experimental analysis,the maximum and minimum values of the average error of each 1 m driving of the transport robot are 0.052 m and 0.036 m,respectively,which can meet the positioning requirements of the transport robot in the orchard.2D laser radar is used for positioning of fruit trees.The average deviation of X axis is 0.176 m and the average deviation of Y axis is 0.146 m by comparing the actual measurement of fruit tree coordinates with the coordinates obtained by the algorithm,which meets the requirements of fruit tree positioning accuracy.(4)Considering the actual environment of the orchard,the orchard navigation path fitting algorithm was studied.A least squares fitting navigation path algorithm based on interior points is proposed,and compared with RANSAC and least squares fitting path algorithms,it is obtained that the difference of average radian of the least squares fitting navigation path algorithm based on interior points is-0.269 and-0.032,respectively,indicating that the least squares fitting navigation path algorithm based on interior points is superior to RANSAC and least squares fitting path algorithm,so this algorithm is more suitable for the fitting of orchard navigation line.(5)Aiming at the control problem of orchard equipment,the decision-making algorithm of orchard control was studied.After obtaining the lateral deviation and heading deviation between the transport robot and the navigation line,the control method of the transport robot based on fuzzy control is proposed according to the size of the transport robot and the width of the fruit tree.The navigation control decision of transport robot is realized.The field test is carried out at the speed of 0.25 m/s,0.50 m/s and 0.75 m/s for the designed transport robot.The maximum deviation of the navigation system is 0.414 m,the average deviation is 0.039 m,and the standard deviation is 0.150 m,which verifies the navigation effect of the transport robot and shows that the navigation system has practical significance.