基于RBF网络的机械手臂逆问题的求解

机械手臂的研究，要涉及到机械手运动学逆问题，即由已知机械手末端的位姿求出机械手各个关节的关节量。逆问题的解析法存在着无解的问题，数值解法计算量太大。应用RBF神经网络良好的非线性逼近能力，对机械手臂的运动学逆问题进行非线性逼近，利用训练好的网络可以精确，快速地求出逆问题的解，且在多解情况下可以求得最优解。

Solution of Manipulator Arm's Reverse Kinematics Based on RBF Net

The research of manipulator arm refers to reverse kinematics, that is resolving the joints＇ variables with position of the manipulator arm＇s terminal. There will be no solution to solve reverse kinematics with analytic method, and more computation with numerical method. This paper resorts to RBF net＇s ability of good nonlinear approximation, and constructions and trains one net to approach reverse kinematics. The net can accurately and speedily solve reverse kinematics,furthermore can get the optimization solution at the case of more solutions.