串联6自由度机器人关节刚度辨识与误差补偿研究

Research on Joint Stiffness Identification and Error Compensation of the Serial Six DOF Robot

为提高串联6自由度机器人的绝对定位精度,针对几何参数误差补偿后的工业机器人关节刚度参数展开研究。首先,基于虚拟关节模型建立了工业机器人一维关节刚度误差模型。其次,为提高关节刚度参数的辨识精度与效率,利用BP神经网络对刚度误差模型进行拟合,以优化遗传算法的初始种群适应度。最后,利用激光跟踪仪AT930和ER10L-C10机器人进行实验,验证以上误差模型与关节刚度参数辨识算法。实验结果表明,经过关节刚度误差补偿后,机器人的平均距离误差与最大距离误差分别为0.2485mm与0.3332mm。相比于补偿前的距离误差,机器人定位精度提高了33.7%。因此,通过改进遗传算法辨识得到的机器人关节刚度参数能够有效地提高机器人定位精度。

To improve the absolute positional accuracy of the serial six-DOF robot,the joint stiffness er ror of industrial robots after geometric parameter error compensation is studied. Firstly,the one-dimensional joint stiffness error model of industrial robots is established based on the virtual joint model. Secondly,in or der to improve the identification accuracy and efficiency of joint stiffness parameters, the BP neural network is applied to fit the stiffness error model to optimize the initial population fitness of genetic algorithm. Finally, the laser tracker AT930 and ER10L-C10 robot are used to verify the above error model and joint stiffness pa rameter identification algorithm. The experimental results show that the average distance error and maximum distance error of the robot are 0.248 5 mm and 0.333 2 mm respectively after the joint stiffness error compen sation. Compared with the distance error before error compensation,the positional accuracy of robot is im proved by 33.7%. Therefore,through the proposed improved genetic algorithm can identify the joint stiffness parameters accurately and improve the positional accuracy effectively.