基于自适应联邦滤波的AGV定位研究  

Research on AGV Positioning Based on Adaptive Federated Filter

作　　者：温尊旺 王尧尧 陈柏[1] 姚佳烽 WEN Zunwang;WANG Yaoyao;CHEN Bai;YAO Jiafeng(College of Mechanical and Electrical Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 210016,China)

机构地区：[1]南京航空航天大学机电学院,江苏南京210016

出　　处：《机械制造与自动化》2022年第5期247-251,256,共6页Machine Building & Automation

基　　金：国家自然科学基金项目(52175097)。

摘　　要：为了提高室内AGV的定位精度,提升其灵活性同时降低定位成本,在Mecanum轮全向机器人平台下,设计一款精度较好的低成本组合导航系统。针对卡尔曼滤波中噪声参数无法准确估计,采用自适应滤波来减小系统的扰动。提出一种多尺度融合方案,解决各传感器更新频率不一致的问题。通过Mecanum轮全向机器人的仿真和轨迹实验表明:所设计的组合导航系统具有较好的精度,最大定位误差减小了69%,方均根误差减小了56.6%,且能够有效抑制扰动,定位性能得到提升。In order to improve the positioning accuracy and flexibility of indoor AGV and reduce the positioning cost, a low-cost integrated navigation system of high accuracy is designed on the Mecanum wheel omnidirectional robot platform. In the light of inaccurate estimation of noise parameters in Kalman filter, an adaptive filter is adopted to reduce the disturbance of the system. A multi-scale fusion scheme is proposed to overcome the inconsistency of update frequency of each sensor. The simulation and trajectory experiments of Mecanum wheeled omnidirectional robot show that the designed integrated navigation system has good accuracy with maximum positioning error reduced by 69% and the root mean square error by 56. 6%, which can effectively suppress disturbance and improve positioning performance.

关 键 词：MECANUM轮 自适应滤波 联邦卡尔曼滤波 组合导航 

分 类 号：TP23[自动化与计算机技术—检测技术与自动化装置]