抓取机械臂基于ART-RBF学习算法的运动学分析
|
摘要
针对在研究抓取机械臂运动学逆解过程中所遇到的困难,计划选择一种基于软竞争机制的ART-RBF模型。在传统RBF神经网络的基础上,自适应控制生成隐含层节点数目,将相似度软竞争使用在第一阶段的学习中。采用软竞争机制,能够将隐含层各个节点都参与到对样本的学习,节点利用率提高,同时也能够减少类间混叠处样本的误差,提高预测精度。最后,利用ADAMS对机械臂进行运动仿真,并与所得运动学正解相比较,用来验证正解方程的正确性。结果表明,该软竞争算法能够一定程度上提高预测精度,仿真结果表明了所得正解数据的准确性,为后续运动控制提供依据。
In view of the difficulties encountered in the study of the kinematics inverse kinematics ofgrasping manipulator, a ARTRBF model based on soft competition mechanism is selected. On the basis of traditional RBF neural network, adaptive control generates the number of hidden layer nodes, and the similarity softcompetition is applied in the first stage of learning. Using the soft competition mechanism, each node of the hidden layer can be involved in the learning of the sample, the utilization rate of the node is improved, and the errorof the sample in the inter class overlap can be reduced, and the prediction accuracy can be improved. Finally,the motion simulation of the manipulator is carried out by ADAMS, and it is compared with the forward kinematics solution to verify the correctness of the positive solution equation. The results show that the soft competitionalgorithm can improve the prediction accuracy to a certain extent. The simulation results show the accuracy ofthe positive solution data and provide the basis for the followup motion control.
In view of the difficulties encountered in the study of the kinematics inverse kinematics ofgrasping manipulator, a ARTRBF model based on soft competition mechanism is selected. On the basis of traditional RBF neural network, adaptive control generates the number of hidden layer nodes, and the similarity softcompetition is applied in the first stage of learning. Using the soft competition mechanism, each node of the hidden layer can be involved in the learning of the sample, the utilization rate of the node is improved, and the errorof the sample in the inter class overlap can be reduced, and the prediction accuracy can be improved. Finally,the motion simulation of the manipulator is carried out by ADAMS, and it is compared with the forward kinematics solution to verify the correctness of the positive solution equation. The results show that the soft competitionalgorithm can improve the prediction accuracy to a certain extent. The simulation results show the accuracy ofthe positive solution data and provide the basis for the followup motion control.
引文
[1]熊有伦,丁汉,刘恩沧.机器人学[M].北京:机械工业出版社,1993:89-93.
[2]DUGULEANA M,BARBUCEANU F G,TEIRELBAR A,et al.Obstacle avoidance of redundant manipulators using neural networks based reinforcement learning[J].Robotics and Computer-Integrated Manu-facturing,2012,28(2):132-146.
[3]POIGNET P,GAUTIER M.Nonlinear model predictive control of a robot manipulator[C]//Proceedings the 6th International Workshop on Advanced Motion Control,March 30-April 1,2000,Nagoya,Japan.New York:IEEE,2000:401-406.
[4]CHIDDARWAR S S,BABU N R.Comparison of RBF and MLPneural networks to solve inverse kinematic problem for 6R serial robot by a fusion approach[J].Engineering Applications of Artifici Intelligence,2010,23(7):1083-1092.
[5]贠超,刘刚,王刚,等.基于RBF神经网络和二次规划的冗余机械臂避障问题研究[J].机电工程,2016,33(1):1-7.
[6]张翠,杨剑锋,张峰.基于EC-RBF学习算法的机械臂运动学逆解研究[J].制造业自动化,2013(14):64-65.
[7]蔡自兴.机器人学[M].北京:清华大学出版社,2000:40-45.
[8]BEZDEK J C.Pattern recognition with fuzzy objective function algorithm[M].New York:Plenum,1981:1-13.
[9]姜鹏飞,蔡之华.基于遗传算法和梯度下降的RBF神经网络组合训练方法[J].计算机应用,2007,27(2):366-368.
[10]李增刚.ADAMS入门详解与实例[M].北京:国防工业出版社,2010:111-120.
[2]DUGULEANA M,BARBUCEANU F G,TEIRELBAR A,et al.Obstacle avoidance of redundant manipulators using neural networks based reinforcement learning[J].Robotics and Computer-Integrated Manu-facturing,2012,28(2):132-146.
[3]POIGNET P,GAUTIER M.Nonlinear model predictive control of a robot manipulator[C]//Proceedings the 6th International Workshop on Advanced Motion Control,March 30-April 1,2000,Nagoya,Japan.New York:IEEE,2000:401-406.
[4]CHIDDARWAR S S,BABU N R.Comparison of RBF and MLPneural networks to solve inverse kinematic problem for 6R serial robot by a fusion approach[J].Engineering Applications of Artifici Intelligence,2010,23(7):1083-1092.
[5]贠超,刘刚,王刚,等.基于RBF神经网络和二次规划的冗余机械臂避障问题研究[J].机电工程,2016,33(1):1-7.
[6]张翠,杨剑锋,张峰.基于EC-RBF学习算法的机械臂运动学逆解研究[J].制造业自动化,2013(14):64-65.
[7]蔡自兴.机器人学[M].北京:清华大学出版社,2000:40-45.
[8]BEZDEK J C.Pattern recognition with fuzzy objective function algorithm[M].New York:Plenum,1981:1-13.
[9]姜鹏飞,蔡之华.基于遗传算法和梯度下降的RBF神经网络组合训练方法[J].计算机应用,2007,27(2):366-368.
[10]李增刚.ADAMS入门详解与实例[M].北京:国防工业出版社,2010:111-120.