摘要
在大型结构件装配过程中,为了减小多激光雷达组网系统的测量误差,并实现装配轨迹的自动修正,设计了基于聚类算法的装配轨迹自动修正系统。通过聚类分析完成对站位布局的优化,利用残差补偿实现多站数据的融合,并完成了5种不同状态的单站及多站距离精度测试。实验结果显示,在单站精度几乎一致的条件下,经优化布局后的系统转站测量不确定度由1.25 mm提升至0.45 mm。在装配轨迹修正实验中,获得了半径为3.5 m圆域内激光雷达随偏转角变化的匹配数据规律。测试分析了5个位置点的坐标值与位姿偏移量,在三个轴向上的位置偏差均值分别为0.042 mm、0.033 mm和0.039 mm,位姿偏角均值分别为0.25°、0.23°和0.49°。可见,装配轨迹修正量与转站测量不确定度相近,该系统可以实现对装配过程的在线自动修正。其可有效降低多激光雷达组网系统的数据融合误差,在自动化装配领域具有一定的价值。
In the assembly process of large-scale structural parts, in order to reduce the measurement error of the multi-lidar networking system and realize the automatic correction of the assembly trajectory, an automatic correction system of the assembly trajectory based on the clustering algorithm is designed.The optimization of the station layout is completed through cluster analysis, the fusion of multi-station data is realized by residual compensation, and the distance accuracy test of single and multi-station in five different states is completed.The experimental results show that the measurement uncertainty of the optimized layout of the system transfer station improves 1.25 mm to 0.45 mm under the condition that the accuracy of the single station is almost the same.In the assembly trajectory correction experiment, the matching data law of the lidar with the deflection angle in a 3.5 m radius circle is obtained.The test analyzes the coordinate values and positional offsets of five position points with mean positional deviations of 0.042 mm, 0.033 mm, and 0.039 mm in the three axes and means positional offsets are 0.25° and 0.23°,respectively.It can be seen that the correction amount of the assembly trajectory is similar to the uncertainty of the transfer station measurement, and that the system can realize the online automatic correction of the assembly process.It can effectively reduce data fusion errors in multi-lidar networking system, and is of value in the field of automated assembly.
作者
李超然
崔志明
徐晶
LI Chao-ran;CUI Zhi-ming;XU Jing(College of Information Engineering,Changchun College of Electronic Technology,Changchun 130061,China;Hithink Yondervision(Beijing)Tech.Co.,Ltd.,Beijing 100088,China;College of Business,Changchun College of Electronic Technology,Changchun 130061,China)
出处
《激光与红外》
CAS
CSCD
北大核心
2023年第2期296-301,共6页
Laser & Infrared
基金
吉林省科技创新基金项目(No.2021103135)资助。
关键词
激光雷达组网
轨迹修正
自动装配
聚类算法
Lidar networking
trajectory correction
automatic assembly
clustering algorithm
