机器人逆标定方法比较研究

Comparison research of robot inverse calibration

对两种基于神经元网络的机器人逆标定方法:把机器人位姿和关节角误差作为前馈神经元网络的输入输出,以及把机器人关节角和对应的关节角偏差作为神经元网络输入输出,进行了比较研究。仿真结果表明前者标定效果更好,且更方便简单。把前一种方法的标定结果与传统的几何参数标定方法进行了比较,证明了该方法的有效性,避免了其他传统标定方法繁琐的建模及参数辨识过程。

Two robot inverse calibration methods based on a feed-forward neural network, the first method adopting robot actual poses and corresponding joint errors as input and output of neural network, the second method adopting joint angles and their corresponding errors as input and output, is compared in this paper. Simulation results show that the former method is more effective, convenient and simpler than the latter. Calibration results of the first method are compared with those obtained by traditional parametric methodologies, and the comparison results verify its effectiveness, furthermore, it avoids the complex modeling and parameters identification of the traditional parametric methodologies.