Analysis of Efficient and Fast Prediction Method for the Kinematics Solution of the Steel Bar Grinding Robot

Aiming at the robotization of grinding process in steel bar finishing process, robot can achieve goal fast, efficient, and accurate online operation, a multi-layer forward propagating deep neural network (DNN) method is proposed to efficiently predict kinematic solution robot. The kinematics model are introduced. Based on method, simulations end position orientation, joint angle given. Three different methods, including SGD + tanh, Nadam ELU, used test DNN calculation results show that combining with ELU function has fastest speed higher accuracy be obtained increase iteration times. Finally, optimizer optimize example. optimization this accelerates convergence rate trajectory prediction error improves prediction. Thus, paper an effective when works online.