基于RBF神经网络的机器人平滑圆弧轨迹规划

Trajectory Planning of Robotic Smooth Arc Based on RBF Neural Network

针对一般圆弧插补算法收敛性和稳定性差,逼近效果不理想的问题,提出基于RBF神经网络的圆弧轨迹规划方式。论文从运动学角度出发,分析求解工业四轴机器人运动学正逆问题,建立其D-H数学模型和空间三维模型。通过圆弧轨迹经过的空间三点,求出空间圆弧的参数方程;在Matlab软件中,仿真RBF神经网络规划的圆弧轨迹曲线,结果表明RBF神经网络所规划轨迹平滑稳定,收敛速度快,逼近误差小,从而证明RBF神经网络对机器人轨迹规划的可行性与必要性。

To solve problem that the convergence and stability of the arc interpolation algorithm is not ideal,this paper proposes a circular trajectory planning method based on RBF neural network.From the point of view of kinematics,the normal inverse kinematics problem of four axis robot is analyzed,and the mathematical model of D-H and 3D model is established.Through the space 3points of circular arc track,the parameter equation of space circular arc is obtained.In the Matlab software,emulating the trajectory of the circular interpolation algorithm based in RBF neural network planning,the results show that the trajectory planning of RBF neural network is smooth and stable,the convergence speed is fast and the approximation error is small.So the feasibility and necessity of RBF neural network for robot trajectory planning are proved.