Neural network method for robot arm of service robot based on D-H model

In this paper, a radial basis function neural network was used to solve the robot arm with 4 DOF of a self-designed service robot. Firstly, the D-H model was established for the manipulator and the analysis of forward kinematics was carried out, and the transition relation between the connecting rod coordinates was obtained. Then, according to this relationship, the sample data from joint space to Cartesian space was trained. Finally, the radial basis function neural network was used to find the inverse solution of the manipulator, based on a large amount number of the sample data, and the mapping relationship from the Cartesian space to the joint space. The method was verified by MATLAB simulation, and it can be used to solve the inverse kinematics of robot arm. It met the design requirements of the service robot in a certain range of errors.