| Agricultural robots are an important branch of the field of robotics.Agricultural robots replacing manual operation has become an important trend in the development of modern agriculture.Forest and orchard robots can reduce labor input,improve labor productivity and product quality,and reduce health and safety risks of employees.Autonomous navigation technology is an important part of agricultural mobile robot research,and it plays an important role in promoting the intellectualization of agricultural machinery and the refinement of agricultural production.In order to improve the intellectualization and autonomy of mobile robots in forest and orchard,this paper studied the path planning and autonomous navigation methods of mobile robots in forest and orchard environment,designed the path search algorithm based on environmental information sampling and the real-time navigation method of high and low frequency dual-source information fusion between lines of forest and orchard based on lidar threedimensional point cloud.The main research contents are as follows:(1)Aiming at the working environment of forest and orchard,the overall technical scheme of mobile robot is designed.Beidou integrated navigation system and three-dimensional Lidar are used to collect the robot’s own pose information and surrounding environment information.Designed the robot autonomous navigation hardware system and completed the organization and construction of the experiment platform;The robot autonomous navigation software system based on ROS system is designed to realize the ring global path planning and autonomous navigation of forest orchard.(2)Aiming at the problems of long search time and low search efficiency of traditional path search algorithms in forest and orchard environment,a path planning method based on local domain PCA was proposed.In the global geometric feature map of Linguoyuan,through sampling the environment information in the local domain of the current node,sampling points in the free zone were selected for principal component analysis,the extension direction of the free zone was taken as the growth direction of the new node,and the growth step was changed by the Angle difference of the growth direction of two adjacent nodes,so as to generate the new node.Finally,the effectiveness and speediness of the path planning algorithm are verified in the simulated map and real forest and orchard environment.Experiments show that compared with the traditional path planning algorithm A* algorithm and two-way RRT* algorithm,the number of path nodes of the algorithm in this chapter is reduced by more than 50%,and the search speed and efficiency are fast,which can meet the global path search requirements of forest and orchard.(3)In order to improve the accuracy and robustness of the navigation system of mobile robot in Forest orchard,a real-time navigation method based on Li DAR three-dimensional point cloud based on high and low frequency dual-source information fusion between forest orchard lines was proposed.Through filtering,downsampling and outlier removal were preprocessed to reduce the interference of irrelevant point clouds.The inter-line navigation lines fitted by Newton interpolation algorithm with high frequency updating and nonlinear support vector machine algorithm with low frequency updating were complementary fused,and the stability of the fused navigation lines was optimized.Finally,a spray robot was used as the experimental platform to test the curvature smoothness and tracking accuracy of the navigation line after fusion.The experiment showed that the average curvature of the navigation line after fusion optimization was0.018m-1,which was 54% lower than the average curvature of the interpolation navigation line and 55% lower than that of the NSVM navigation line,which was more beneficial to the trajectory tracking of the robot.To meet the needs of independent navigation between forest and orchard lines. |
PDF Full Text Request