基于RBF神经网络间接求取运动学逆解的研究

Research on Indirect Solution of Inverse Kinematics Based on RBF Neural Network

七自由度工业机器人的几何结构大多满足Pieper准则,所以针对七自由度的封闭解法具有很大的发展空间。提出了一种基于RBF神经网络间接求取运动学逆解的方法,将运动学方程转化成了优化控制问题。采用遗传算法与最佳柔顺性准则相结合的方法,为RBF神经网络算法提供了精确的样本;为了提高神经网络算法的收敛速度以及收敛精度,进行间接求取的方式,引入连杆三角形夹角的概念;为了验证结果的可靠性,以七自由度冗余机械臂为对象,开展了基于RBF神经网络算法间接求逆的优化实验,并对比传统的RBF神经网络求取运动学逆解算法,结果表明,该算法结构简单,且能够显著提高收敛精度和收敛速度。

The geometric structure of the seven-degree-of-freedom(7-DOF)industrial robot mostly meets the Pieper criterion,so the closed solution for the 7-DOF has a lot of room for development.A method based on Radial Basis Function(RBF)neural network to indirectly obtain kinematics inverse solution is proposed.The kinematics equation was transformed into the optimal control problem.The method of combining the genetic algorithm with the best compliance criterion was adopted to provide an accurate sample for the RBF neural network algorithm.In order to improve the convergence speed and the convergence accuracy of the neural network algorithm,an indirect method was adopted to introduce the concept of angle of the connecting rod triangle.In order to verify the reliability of the results,an optimization experiment based on the RBF neural network algorithm for indirect inverse was carried out on a 7-DOF redundant manipulator,and the inverse RBF neural network was compared with the traditional RBF neural network to solve the inverse kinematics algorithm.The results show that the algorithm is simple in structure and can significantly improve convergence accuracy and convergence speed.