The Trajectory Planning On Four-axis Robot Arm Based On RBF Neural Network

Industrial robots are highly integrated intelligent digital products, which plays a crucial role in field of today’s automatic control. In the ever-changing market demand, in order to meet better the needs of industrial production, industrial robot should have a very wide versatility, unique flexibility, higher accuracy and better motion performance. Robot trajectory planning for robot motion control is essential. During movement of the robot to complete the task, the trajectory generation algorithm will affect the stability of the system. Using reasonable path planning algorithm, the robot can be controlled quickly, smoothly and accurately arrived at the designated location, complete the space complex curved path movement. Therefore, the robot trajectory generation algorithm is very worthy of study.Usually using trajectory planning ways in the joint space and Cartesian coordinate space to analyse the robotic trajectory parameters to calculate and generate trajectory in real time. But the traditional way at the time of increasing the degree of freedom of the robot, which effect is very poor because of A lot of calculation. Currently, genetic algorithm and BP neural network are often used to optimize trajectory planning problems, but "premature"and computationally intensive of the genetic algorithm, global approach, slow learning of network BP and other shortcomings, in practical application, the effect is not ideal. Because the fast convergence and high-precision approach capability of RBF neural network, it is suitable for application in trajectory planning of the robot.This paper mainly discusses the four-axis industrial robot arm trajectory planning approach in the workspace.in order to improve speed and accuracy of the robotic arm on the path point fit, using RBF neural network to improve machine arm system performance, the focus of this article focuses on improving the speed and accuracy for the robotic arm to complete job tasks.In the research,firstly analysis of the system structure of four-axis industrial robotic arm, using knowledge of industrial robotics to establish Four-axis manipulator motion model, with Denavit-Hartenberg parameter method got its kinematics equation.secondly, studing structure of the machine arm by MATLAB software with powerful numerical calculation function,get its three-dimensional model,verified the correctness of the model and obtain different coordinate, speed, acceleration of machine arm which has important guiding significance for exploring further robot arm trajectory generation algorithm;lastly,analysis of the trajectory planning algorithm of robotic arm in joint space and Cartesian coordinate space, using RBF neural network fitting and approximation function to Improve machine system performance Designing radial basis function neural network with MATLAB software, simulation of its ability to fit interpolation points.The results show that compared with the traditional trajectory planning approach,the RBF neural network planning trajectory has fast calculation speed, good fitting effect and can optimize fitting approximation ability for path interpolation point, also verified the effectiveness and feasibility of using RBF neural network to plan the robot arm trajectory.