基于径向基神经网络的Delta机器人位置精度补偿  

Position Accuracy Compensation of Delta Robot Based on RBF Neural Network Composites

作　　者：董慧芬[1] 高爽笑 宋金海 DONG Hui-fen;GAO Shuang-xiao;SONG Jin-hai(Robotics Institute, Civil Aviation University of China, Tianjin 300300, China)

机构地区：[1]中国民航大学机器人研究所,天津300300

出　　处：《科学技术与工程》2020年第31期12883-12889,共7页Science Technology and Engineering

基　　金：天津市自然科学基金(17JCYBJC18200)。

摘　　要：针对Delta并联机器人末端控制精度问题,提出一种基于径向基神经网络的提高Delta并联机构运动学控制精度的方法。首先对Delta并联机器人的运动学逆解进行分析,探讨了影响控制精度的因素和现有提高控制精度方法的局限性。其次,求解Delta并联机器人的工作空间,结合实际工作,通过试验采集训练样本。以末端实际位置为输入样本,末端的期望位置与实际位置之差为输出样本,进行径向基神经网络模型训练,得到末端实际位置与位置偏差之间的非线性映射关系,基于此设计位置补偿策略。最后,在Delta机器人平台上进行实验验证,使用训练好的径向基神经网络网络结合运动学逆解,对Delta机器人末端进行轨迹跟踪控制。实验结果表明,末端控制误差由±30 mm减小到±5 mm,有效地减少了末端位置误差,为Delta机器人精准控制提供了一种简单易行的方法。Aiming at the end control accuracy of Delta parallel robot,a method based on radial basis function network(RBF)to improve the kinematics control accuracy of delta parallel mechanism was proposed.Firstly,the inverse kinematics solution of Delta parallel robot was analyzed,and the factors affecting the control accuracy and the limitations of the existing methods to improve the control precision were discussed.Secondly,the working space of Delta parallel robot was solved,and the training samples were collected by experiment.Taking the actual position of the end as the input sample and the difference between the expected position and the actual position of the end as the output sample,the RBF neural network model was trained to obtain the nonlinear mapping relationship between the actual position and the position deviation of the end.And based on this,location compensation strategy was designed.Finally,experimental verification was carried out on the Delta robot platform,and trajectory tracking control was carried out on the end of the Delta robot by using the trained RBF network combined with the inverse kinematic solution.The results show that the end control error is reduced from±30 mm to±5 mm,which effectively reduces the end position error and provides an simple and easy way for Delta robot precise control.

关 键 词：Delta机器人 径向基神经网络 非线性 误差分析 误差补偿 

分 类 号：TP242.2[自动化与计算机技术—检测技术与自动化装置]