Research On Environment Construction And Path Planning Of Autonomous Vehicles For Pepper Harvesting

Autonomous harvesting vehicles can greatly reduce the labor intensity of agricultural producers,so autonomous harvesting vehicles are an important development direction in the field of agricultural machinery in the future.Environmental construction and path planning technology are the key technologies for autonomous harvesting vehicles.Environmental construction can enable vehicles to understand the surrounding environmental information so as to complete the follow-up work.Path planning allows vehicles to successfully avoid obstacles and complete harvesting operations safely and efficiently.In this paper,environmental modeling and path planning are studied for pepper field scenarios.(1)Combined with the pepper field scene,the visual SLAM system is optimized by adding line features,and constraint conditions are added in the feature matching process to improve the quality of feature matching.The pose estimation model of the optimized system was constructed,and solve the Jacoby matrix in the mathematical model.Through experiments,the accuracy of feature extraction,matching and pose estimation is verified.Finally,3D raster map is used to show the effect of map construction.(2)Two commonly used path planning algorithms are combined to make up for the shortcomings of their respective algorithms.According to the actual pepper field scene,the appropriate driving method and turning mode in the full coverage path planning algorithm are selected.The artificial potential field method suitable for pepper field was selected as the local path planning algorithm,and the problems of unreachable target and local minimum value were solved.Through experiments,the feasibility of covering path planning algorithm and artificial potential field method is verified respectively.(3)The optimized visual SLAM system is tested through the actual pepper field scene,and the experiment proves that the system still meets the requirements of real-time performance;the accuracy of the system has been greatly improved,the accuracy has been improved by more than 14.6%;three-dimensional raster map consumes very little memory,which is beneficial to improve the efficiency of the system.The experimental scenario of path planning is built by simulating the pepper field.While the vehicle tracks the full coverage path,it can also complete the local path planning,and the error is within the acceptable range,which proves that the path planning algorithm proposed in this paper is feasible.