基于GWO-ELM算法与模糊控制的无标定视觉伺服研究

Research on Uncalibrated Visual Servo Based on GWO-ELM Algorithm and Fuzzy Control

针对传统基于图像的视觉伺服系统运行速度慢,图像雅可比矩阵的求解受标定精度影响的问题,提出一种基于灰狼算法优化极限学习机(GWO-ELM)与模糊控制相结合的视觉伺服控制方法。该方法利用灰狼算法(GWO)优化ELM模型初始权重增加模型稳定性,估计图像雅可比矩阵伪逆预测机械臂末端运动速度,之后引入模糊控制(Fuzzy Control)设计视觉伺服控制器构建无标定视觉伺服控制系统,并进行上机实验。实验结果表明,Fuzzy Control-GWO-ELM-IBVS的运行效率相对于GWO-ELM-IBVS得到了提升,定位误差能控制在规定阈值,验证了提出的无标定视觉伺服控制系统的有效性。

To address the problems of slow operation of traditional image-based visual servo system and the impact of calibration accuracy on the solution of image Jacobi matrix,this paper proposes a visual servo control method based on the combination of gray wolf optimized extreme learning machine(GWO-ELM)and fuzzy control.The method uses the gray wolf algorithm(GWO)to optimize the initial weights of the ELM model to increase the stability of the model,estimates the image Jacobi matrix pseudo-inverse to predict the end motion speed of the robot arm,and then introduces the fuzzy control(Fuzzy Control)to design the visual servo controller to build a calibration-free visual servo control system and conducts the experiments on the machine.The experimental results show that the operating efficiency of Fuzzy Control-GWO-ELM-IBVS is improved compared with that of GWO-ELM-IBVS,positioning errors can be controlled within specified thresholds,which verifies the effectiveness of the calibration-free visual servo control system proposed in this thesis.