Intelligent agricultural machine static and dynamic object detection path planning method based on convolutional neural network

Abstract

The invention relates to a path planning method for intelligent agricultural machine static and dynamic object detection based on a convolutional neural network, which is used for solving the problem that the agricultural machine path planning method is poor in effectiveness and stability due to the fact that the classification efficiency is low and obstacles cannot be effectively distinguished in the agricultural machine path planning process, and comprises a training obstacle classification module; a global map is constructed, and global path planning is completed; the agricultural machine runs according to the global path planning scheme, obstacles on the running path are detected in real time, when the obstacles are detected, the obstacle classification module which completes training is used for conducting dynamic and static classification on the detected obstacles, and instant local obstacle avoidance is conducted according to the classification result till the agricultural machine runs on the whole path. The point-by-point label matrix output by the CNN is changed into the instance label of the scanned object by using the K-means clustering algorithm, the resolution efficiency and the resolution accuracy are improved, the obstacles are subjected to dynamic and static classification, different obstacle avoidance methods are adopted, and the safety and the stability are higher.

Description

technical field [0001] The invention relates to a path planning method for static and dynamic object detection of intelligent agricultural machinery based on a convolutional neural network, in particular to a method that functions when the original global path is blocked by objects, and belongs to the field of intelligent agricultural machinery equipment. Background technique [0002] The development of modern agriculture has made the path planning of agricultural machinery a research hotspot, and laser radar technology is involved in the path planning of agricultural machinery. In recent years, with the rise of LiDAR technology, 3D point cloud data has been widely used in remote sensing mapping and other fields. Point cloud data is not affected by factors such as lighting and shadows, and has rich three-dimensional spatial information. At present, point cloud processing technology generally applies machine learning to point cloud data classification. Among them, Convoluti...

AI Technical Summary

Problems solved by technology

[0005] In order to solve the problem of poor effectiveness and stability of the agricultural machinery path planning method due to the low classification efficiency and the inability to effectively distinguish obstacles in the path planning process of agricultural machinery, the present invention proposes a static, Path Planning Method for Dynamic Object Detection

Images