Robot autonomous reasoning and scene understanding method for orchard tree shape

Abstract

The invention provides a robot's orchard tree frame type autonomous reasoning and scene understanding method, which belongs to the field of agricultural robots. For the complex environment of semi-structured orchards, the robot autonomously navigates and locks the canopy through laser radar, obtains the main branch-column skeleton features of the canopy, compares it with the orchard agronomic knowledge base built in the control system, and realizes the orchard tree shape Based on the independent reasoning of different orchard tree shapes, it automatically selects the corresponding initial positioning mode of the hand camera, completes the rough three-dimensional contour imaging of the canopy, and finally automatically selects the corresponding canopy partition mode to realize close-range scene understanding, target recognition and positioning. The present invention solves the problem of poor adaptability and extremely low efficiency of completely relying on exploratory detection of unknown orchard scenes when the robot works, and realizes autonomous and continuous understanding of orchard scenes.

Description

technical field [0001] The invention relates to the field of agricultural robots, in particular to a method for autonomous reasoning and scene understanding of a robot's orchard tree frame. Background technique [0002] The orchard is a typical semi-structured complex environment, and the effective understanding of the orchard scene is the core to ensure the autonomous continuous operation of the robot. Based on a full understanding of the orchard scene, the robot can navigate in the orchard and make decisions about complex actions such as continuous fruit picking, bagging, pruning, leaf threshing, and pollination. However, the existing research is still difficult to meet the needs of continuous operation of the robot in the orchard site, the main problems are as follows: [0003] (1) There are great differences in orchard construction specifications and tree frame types. Existing navigation and operations either rely on prior information and pre-navigation mapping, or comp...

AI Technical Summary

Problems solved by technology

However, the existing research is still difficult to meet the needs of continuous operation of the robot in the orchard site, the main problems are as follows:

[0003] (1) There are great differences in orchard construction specifications and tree frame types. Existing navigation and operations either rely on prior information and pre-navigation mapping, or completely rely on the robot's exploratory detection of completely unknown and complex orchard and tree crown environments , the huge computing power consumption leads to poor adaptability and extremely low efficiency of the robot;

[0004] (2) Existing research mainly focuses on single research on tree row detection and fruit segmentation and extraction, but lacks the autonomous and continuous understanding of different scales and different tasks in the continuous operation process of robots in orchards, which cannot meet the needs of robots walking in orchards, Complex decision-making needs for finding goals and jobs

Images