基于视觉融合与GNSS的苹果采摘机器人自主导航系统设计与试验OA
Design and experiment of autonomous navigation system for apple picking robot based on visual fusion and GNSS
中国作为全球主要的苹果生产国,面临着劳动力成本上升和人口结构变化所带来的农业用工紧张问题,推动果园作业向自动化、智能化转型已成为提升农业效率的关键因素.针对标准化栽培模式下的苹果果园环境,设计并实现一种基于全球导航卫星系统(GNSS)与视觉融合导航的自动导航系统.该系统采用阿克曼轮式底盘,集成前轮转角检测与自动转向控制模块.为提升系统在复杂果园场景中的适应能力,提出了一种基于点云数据、GNSS精度与路径偏差等关键参数的导航模式动态切换策略,实现视觉导航与GNSS导航之间的动态转换,使其能够满足不同作业区域对导航精度的需求.在路径规划方面,构建了结合梭式套行模式与三阶贝塞尔曲线的全覆盖导航路径:行间路径通过双目视觉点云的滤波、下采样与线性拟合提取导航中心线,地头路径通过果园几何参数生成贝塞尔曲线轨迹,以优化换行转弯路径,减少冗余行驶距离.田间试验结果显示,系统在不同行驶速度下均实现高精度路径跟踪,最小横向偏差为0.052 m;地头换行中跨两行与跨一行导航平均偏差分别为0.052 m与0.066 m,满足实际作业精度要求.综上所述,研究提出的导航系统具有良好的环境适应性与导航精度,为果园自动化作业装备的工程化应用提供有力支撑.
As one of the world leading apple producers,China is facing increasing agricultural labor shortages caused by rising labor costs and demographic shifts.Promoting the automation and intelligent transformation of orchard operations has become a key factor in improving agricultural efficiency.In response to the standardized cultivation patterns of modern apple orchards,this study designs and implements an autonomous navigation system based on the fusion of Global Navigation Satellite System(GNSS)and vision-based navigation.The system is built on an Ackermann-steering chassis and integrates a front-wheel angle detection module and automatic steering control.To enhance the system adaptability in complex orchard environments,a dynamic navigation mode switching strategy is proposed,which relies on key parameters such as point cloud data quality,GNSS positioning accuracy,and path deviation.This enables intelligent switching between GNSS and vision-based navigation modes,allowing the system to meet varying accuracy requirements across different operational areas.In terms of path planning,a full-coverage navigation strategy is developed by combining a shuttle-style inter-row pattern with third-order Bézier curves.Inter-row paths are extracted by filtering,down sampling,and linearly fitting the stereo vision point cloud to determine the navigation centerline.Headland turning paths are generated based on orchard geometry parameters using Bézier curves,optimizing turning maneuvers and reducing unnecessary travel.Field tests demonstrate that the system achieves high-precision path tracking at various travel speeds,with a minimum lateral deviation of 0.052 meters.During headland turning,the average navigation deviations for two-row and single-row crossings are 0.052 meters and 0.066 meters,respectively,meeting the accuracy requirements of practical orchard operations.In conclusion,the proposed navigation system demonstrates strong environmental adaptability and navigation precision,providing solid technical support for the engineering application of automated orchard equipment.
问仕威;葛亚豪;贾炯龙;卫乃硕;陈雨;陈军
西北农林科技大学机械与电子工程学院,陕西 杨凌,712100西北农林科技大学机械与电子工程学院,陕西 杨凌,712100西北农林科技大学机械与电子工程学院,陕西 杨凌,712100西北农林科技大学机械与电子工程学院,陕西 杨凌,712100西北农林科技大学机械与电子工程学院,陕西 杨凌,712100||陕西省农业装备工程技术研究中心,陕西 杨凌,712100西北农林科技大学机械与电子工程学院,陕西 杨凌,712100||陕西省农业装备工程技术研究中心,陕西 杨凌,712100
农业科技
视觉融合导航梭式套行全局路径规划苹果采摘机器人标准化果园导航模式动态切换策略
vision-based fusion navigationshuttle coverage global path planningapple harvesting robotstandardized orcharddynamic navigation mode switching strategy
《智能化农业装备学报（中英文）》 2026 (1)
31-42,12
陕西省科学技术厅项目(32272001)国家重点研发计划项目子课题(2022YFD1900304) Shaanxi Provincial Department of Science and Technology Project(32272001)Sub-project of the National Key R&D Program of China(2022YFD1900304)
评论