机械手轨迹规划中的BP神经网络方法及仿真

The Method and Simulation of BP Neural Network in Manipulator's Trajectory Planning

机械手轨迹规划就是控制机械手使之精确地跟踪目标轨迹。文章是在关节空间中进行轨迹规划,将路径点映射成各关节变量角度值,然后对每个关节拟合一个时间函数,此函数往往是高阶的、非线性的。而BP神经网络可以实现从输入到输出的任意非线性映射,能够通过网络学习训练达到逼近非线性函数的目的。文章先建立了一个BP神经网络模型,然后通过Matlab进行仿真,验证了BP神经网络对机械手的非线性运动轨迹的实时性控制的可行性。

Manipulator trajectory planning is to control the manipulator in order to accurately follow the target track. This paper plans the path in the joint trajectory space. Firstly the path is mapped into the joint variable values, and then every joint is imitated into a time function. This function is always high-order and non- linear. But BP neural network can be achieved from the input to the output of arbitrary nonlinear mappingo through network learning and training to achieve the nonlinear function. This paper establishes a BP neural network model, and carries on the simulation through Matlab. The result confirms that the BP neural network can keep the control of the manipulator＇ s non-linear trajectory planning in real time.