摘要
以8自由度凿岩机械臂为研究对象,提出了一种基于运动轨迹的求解多关节冗余机器人运动学逆解方法。首先构建求解机械臂逆运动学问题的BP神经网络模型,然后采用贝叶斯方法,把机械臂各关节在运动轨迹上的转动或移动变化总和作为优化函数的另一个约束项,建立求解逆运动学的贝叶斯-BP网络模型,进行仿真试验。仿真试验证明:构造的贝叶斯-BP神经网络模型可克服BP网络无反馈连接,易陷于局部最少及训练次数少的缺点;该方法在求解位于连续轨迹上的多个工作点的逆运动学问题时,求解的机械臂各关节转动或移动变化平缓,而且逆运动学求解精度可满足控制要求。
For studying the inverse kinematics of an 8-DOF drilling robot, a Bayesian-BP neural network model is presented to solve the multi-joint redundant robot inverse kinematics in a continuous path. First, a BP neural net- work model is constructed to solve the problem of robot inverse kinematics, and then the Bayzain method is adopted to construct the optimization function, in which the influence of the sum rotation or moving changes per joint is in- cluded. Finally the Bayesian-BP network model is built for the inverse kinematics of an 8-DOF drilling robot. The simulation shows that the Bayesian-BP neural network model can overcome the shortcomings of a BP neural net- work, such as no feedback or more training numbers. Rotations or moving changes per joint are smooth in the multiple working points of the robot continuous path, and the calculation precision can meet with the requirement of real time control.
出处
《机械科学与技术》
CSCD
北大核心
2009年第7期862-866,共5页
Mechanical Science and Technology for Aerospace Engineering
基金
湖南省自然科学基金项目(06JJ5099)资助
