摘要
为了提高固定双目相机手眼标定精度,文章提出了一种利用遗传算法优化的反向传播(back-propagation,BP)神经网络的手眼系统标定方法。分别将靶标在相机坐标系与机械臂坐标系下的坐标作为输入量与输出量建立BP神经网络标定模型,并针对BP神经网络模型存在易陷入局部极小值、收敛速度慢的缺点,引入遗传算法来优化BP神经网络的初始权值和阈值。实验结果表明:与BP神经网络模型和传统的多元线性回归拟合模型相比,经遗传算法优化后的BP神经网络模型能显著降低标定误差,加快收敛速度;遗传BP神经网络模型样本预测结果与实测值之间的距离平均误差为1.7126 mm,优于BP神经网络模型的平均误差2.4002 mm和线性回归模型的平均误差3.3771 mm。
In order to improve the accuracy of fixed binocular camera hand-eye calibration,a hand-eye system calibration method using back-propagation(BP)neural network optimized by genetic algorithm was proposed.The BP neural network calibration model was established by using the coordinates of the target in the camera coordinate system and the robot arm coordinate system as the input and output.In view of the disadvantages of the BP neural network model,such as being trapped in the local minimum and slow convergence speed,a genetic algorithm was introduced to optimize the initial weights and thresholds of the BP neural network.The results show that compared with the traditional multiple linear regression fitting model and BP neural network model,the BP neural network model optimized by genetic algorithm can significantly reduce the calibration error and speed up the convergence.The average errors for the measured distances between the tested results and the predicted ones by the proposed genetic BP neural network model,the conventional BP neural network model,and the linear regression fitting model were 1.7126 mm,2.4002 mm,and 3.3771 mm,respectively.
作者
丁雷鸣
徐海明
吕品
赵丹阳
严亚飞
孙丙宇
DING Leiming;XU Haiming;LYU Pin;ZHAO Danyang;YAN Yafei;SUN Bingyu(Institute of Industry and Equipment Technology, Hefei University of Technology, Hefei 230009, China;Institute of Intelligent Machines, Hefei Institutes of Physical Science, Chinese Academy of Sciences, Hefei 230031, China;Tai’an Hualu Metalforming Machine Tool Co., Ltd., Tai’an 271000, China)
出处
《合肥工业大学学报（自然科学版）》
CAS
北大核心
2020年第9期1159-1163,共5页
Journal of Hefei University of Technology：Natural Science
基金
安徽省科技重大专项资助项目(17030901104)。
