基于RBF和BP网络的机器人逆运动学求解

针对传统的求逆运动学方法相当复杂以及一般的神经网络收敛速度慢、精度不高的缺陷,提出一种由1个RBF(Radial Basis Function)网络和2个BP(Back Propagation)网络组成的系统来解决运动学逆问题,输入数据分别通过3个并行的神经网络,对输出分别求正运动学解,计算误差,选择误差最小的作为系统的输出,其中BP网络运用LM(Levenberg-Marquardt)方法进行训练.仿真表明:该方法可以有效的解决运动学逆问题,避免了传统解法中的一些棘手问题.

The solution to inverse kinematics of robot based on RBF and BP neural networks

Many traditional solutions to inverse kinematics are usually complex and general neural networks have slow convergence velocity and low precision.A neural network for inverse kinematics solution approach has been presented.The structure of the proposed method is based on one RBF（Radial Basis Function） and two BP（Back Propagation） networks.At the end of parallel implementation,the results of each network are evaluated using direct kinematics equations to obtain the network with best result.The BP networks use Levenberg-Marquardt training algorithm.Simulations show that inverse kinematics problem can be solved effectively and many intractable problems in traditional way can be avoided.