视觉反馈机械手臂的控制研究

发布时间：2018-05-14 18:18

本文选题：视觉反馈 + 迭代学习控制　；参考：《西南科技大学》2017年硕士论文

【摘要】：针对进行重复作业的机械手臂控制系统,传统的基于系统模型的算法已很难处理系统参数不确定性和拒绝外部干扰,难以让机械手臂的轨迹跟踪误差迅速收敛。通过视觉获取的信息反馈到控制系统以提升机械手臂轨迹跟踪精度,分析了图像处理中的目标定位与坐标系统的转换,为重复控制系统的期望轨迹提供了依据。通过对视觉反馈机械手臂控制算法学习研究后,本文采用基于增益矩阵估计的迭代学习算法来提高机械手臂在重复任务中的轨迹跟踪精度。本文分析了机械手臂的正、逆运动学和动力学,通过对不同自由度的机械手臂仿真和不同增益矩阵的迭代学习控制算法的轨迹跟踪误差收敛速度对比分析,研究了控制系统对重复轨迹的跟踪精度问题。针对PD型迭代学习控制轨迹跟踪误差收敛速度不理想问题,提出基于径向基函数神经网络迭代学习控制优化控制律的增益参数。仿真对比PD型迭代学习控制的轨迹跟踪的误差收敛速度,结果表明优化后的控制算法提升了机械手臂轨迹跟踪误差收敛速度。针对不同的增益矩阵对轨迹跟踪控制误差收敛的影响,采用了控制系统频域特征方程求解方法,提出了将重复控制器中增益矩阵转换成倾斜增益矩阵以提高系统的误差收敛速度。仿真对比基于径向基函数神经网络迭代学习控制的轨迹跟踪误差的收敛速度,结果表明基于倾斜矩阵的控制系统提升了机械手臂轨迹跟踪误差收敛速度,并为控制系统的轨迹跟踪精度分析提供了以提升误差收敛速度的实际增益矩阵数据。本文通过完成优化的迭代学习控制器设计,提高了机械手臂跟踪重复轨迹的误差收敛性能。在上述研究的基础上,在可以与实际机械手臂无缝衔接的ROS操作系统与V-REP软件平台模拟,最后基于VS软件平台实现。
[Abstract]:The traditional algorithm based on system model is difficult to deal with the uncertainty of system parameters and reject external interference, and it is difficult to make the trajectory tracking error of robot arm converge rapidly. The information obtained by vision is fed back to the control system to improve the tracking accuracy of the robot arm. The transformation between target location and coordinate system in image processing is analyzed, which provides the basis for the desired trajectory of the repetitive control system. After studying the control algorithm of the manipulator with visual feedback, an iterative learning algorithm based on gain matrix estimation is proposed to improve the tracking accuracy of the manipulator in repeated tasks. In this paper, the forward and inverse kinematics and dynamics of the manipulator are analyzed. The convergence rate of trajectory tracking error is compared and analyzed through the simulation of the manipulator with different degrees of freedom and the iterative learning control algorithm with different gain matrices. The tracking accuracy of the control system to the repetitive trajectory is studied. Aiming at the problem that the convergence rate of trajectory tracking error of PD iterative learning control is not ideal, the gain parameters of optimal control law based on radial basis function neural network iterative learning control are proposed. The error convergence rate of PD iterative learning control is compared with that of PD type iterative learning control. The results show that the optimized control algorithm improves the convergence speed of trajectory tracking error of manipulator. Aiming at the influence of different gain matrices on the convergence of trajectory tracking control error, the method of solving the characteristic equation in frequency domain of the control system is adopted. In order to improve the error convergence rate of the system, the gain matrix in the repetitive controller is transformed into the tilted gain matrix. The convergence speed of trajectory tracking error based on radial basis function neural network iterative learning control is compared by simulation. The results show that the control system based on tilting matrix can improve the convergence speed of trajectory tracking error of robot arm. The actual gain matrix data to improve the convergence rate of the error are provided for the trajectory tracking accuracy analysis of the control system. In this paper, the optimal iterative learning controller is designed to improve the performance of error convergence in tracking repetitive trajectory of manipulator. On the basis of the above research, the simulation of ROS operating system and V-REP software platform, which can be seamlessly connected with the actual mechanical arm, is implemented based on vs software platform.
【学位授予单位】：西南科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP241

【相似文献】