首页|采用静态数据增强的AGV定位与姿态修正研究

采用静态数据增强的AGV定位与姿态修正研究

扫码查看
搭载激光雷达的移动机器人(AGV)被大量应用于智能工厂的工件运输,但受障碍物和移动物体的影响,AGV在车间内位姿辨别能力较弱。为了解决定位和姿态识别问题,提出一种全新的点云数据融合位姿辨别策略,利用高精度的静态全站仪整体数据融合激光雷达局部、低精度数据,并提出向量权重匹配法来完成AGV的室内定位。设计一种抽样网格卷积法实现异构数据的快速初步定位;建立自适应搜索全站仪数据的基准区域,将它映射到激光雷达数据的对应区域;最后通过向量权重匹配获取AVG的位姿参数。上述方法在6m×8m室内空间进行实验。结果表明:所提方法可达到±7 mm的定位精度与±1。4°的姿态控制识别精度,且能准确补偿激光雷达的扫描误差,提高AGV的位姿识别能力。
Research on AGV Positioning and Posture Correction Using Static Data Enhancement
Automated guided vehicle(AGV)equipped with laser radar are widely used in the transportation of workpieces in intel-ligent factories.However,due to the influence of obstacles and moving objects,AGV has a weak ability to distinguish the position and posture in the workshop.In order to solve the problem of positioning and posture recognition,a new point cloud data fusion position and posture recognition strategy was proposed,using the high-precision static total station data to fuse the local and low-precision data of the laser radar,and the vector weight matching method was proposed to complete the indoor positioning of the AGV.A sampling grid convolution method was designed to realize the rapid preliminary location of heterogeneous data;the reference area for adaptive search of total station data was established and mapped to the corresponding area of laser radar data;finally,the pose parameters of AVG were obtained by vector weight matching.The above method was tested in 6 m×8 m indoor space.The results show that the positioning accura-cy of±7 mm and the posture control recognition accuracy of±1.4° can be achieved,and the scanning error of laser radar can be accu-rately compensated,and the position and posture recognition ability of AGV can be improved.

AGVdata fusionposture recognitionweight matchingpoint cloud matching algorithm

翁润庭、张春良、岳夏、李子涵、龙尚斌、郑仲之

展开 >

广州大学机械与电气工程学院,广东广州 510006

AGV 数据融合 位姿识别 权重匹配 点云匹配算法

国家自然科学基金青年科学基金国家自然科学基金面上项目

5220509052275097

2024

机床与液压
中国机械工程学会 广州机械科学研究院有限公司

机床与液压

CSTPCD北大核心
影响因子:0.32
ISSN:1001-3881
年,卷(期):2024.52(9)