基于遗传优化RBF神经网络的机器人无标定视觉伺服控制

Robot Uncalibrated Visual Servo Control Based on Genetic Optimized RBF Neural Network

为了解决机器人无标定视觉伺服控制系统中精度不高和过程繁琐的问题,提出了基于遗传算法优化的RBF神经网络,对图像特征变化速度和机器人关节角变化速度之间的非线性关系的学习,拟合出两者间的非线性视觉映射关系,从而实现无标定视觉伺服控制。通过对基于图像的眼在手上无标定视觉伺服系统仿真模型进行验证,结果表明,与传统的基于图像的逆雅可比矩阵控制方法相比,文中方法具有更好的精度与效率。

In order to solve the problem of low precision and complicated process in the robot uncalibrated visual servo control system,a RBF neural network based on genetic algorithm optimization is proposed.The nonlinear relationship between the image feature change speed and the robot joint angle change speed is studied,and the nonlinear visual mapping relationship between the two is fitted,so as to realize the uncalibrated visual servo control.The simulation model of an image-based eye-in-hand uncalibrat⁃ed visual servo system is validated.The results show that the proposed method has better accuracy and efficiency than the traditional image-based inverse Jacobian matrix control method.