基于机器视觉的红花采摘机器人自主导航方法研究

打开文本图片集

中图分类号：S244.29 文献标识码：A 文章编号：2095-5553（2026）02-0121-09

Abstract：Aiming atthe saffower picking robot's low precision of navigation line extraction and slow running speed caused bytheifluence of plant growth disorder and complex fieldenvironment，a machinevision-basedautonomous navigation method isproposedtoimprove theautonomous walking precisionand eficiencyofthepickingrobot.Firstly，themonocular camera wasused tocolectthecroprow informationin frontofthe robot，the image was grayscale processedby exceed-green features，and the SUSAN corner point method wasused to extract the feature points after binarization and morphological filtering，andthesaffowercroprow feature points were clusteredbythe improved K一means clustering algorithm，and the least squares method wasused toftthecroprow linesto extract the navigation lines.Secondly，the improved pure pursuit controlalgorithm wasused totrackthenavigationline tocompletethenavigationoperation process.The model was built by usingMATLABforalgorithmsimulation.Theexperimentalresultsshowed thattheoverall navigation lineextraction accuracy of the saffron picking robot was 95.8% ，the average processing time is 68.2ms ，and the navigation line could be extracted accurately. In the test field setting，the average tracking error of linear navigation is 3.32cm ，and the average tracking error of curve navigation is 5.18cm . The autonomous navigation method based on machine vision proposed in this paper can quickly and efectively extract navigation lines，with high accuracy and good navigation effect.

Keywords：saflower picking；autonomous navigation in the field;machine vision;crop row detection;path tracing

0 引言

红花属菊科一年或两年生草本植物，是一种特殊的经济作物，在医药、食品、化工等领域具有广阔的应用前景。（剩余13788字）