基于图像的视觉伺服机械臂控制系统时滞补偿及优化研究

Research on time-delay compensation and optimization for the inner and outer loop control of image-based visual servo manipulator will be expanded in this project for the precision of manipulator control is affected by time delay. For the outer loop control, a method will be proposed for the elimination of the cross-correlation between the process noise and observation noise in the object motion process, then a Kalman colored noise filter will be constructed to estimate the object image with time-delay compensation, here the Kalman filtering steps will be optimized by optimizing the sampling and execution time of the system, the object image will be estimated with time-delay compensation and merged by considering the actual motion characteristic of the object movement, thus the control precision of the outer loop control system will be enhanced for the object image and the image Jacobian matrix estimation is enhanced. For the inner loop control, the kinetic model with time-delay feedback for manipulator will be constructed, for its infinite dimensional, nonlinear, joint periodic motion problem, through linearizing the model and combining lowpass-filter continuous time approximation, the model can be converted to a finite-dimensional feedback dynamics model with time delay is not noticeable, through establishing the mapping matrix between the former discrete time state and the later discrete time state, and seeking the relationship between the mapping matrix and the feedback gain, the feedback gain space can be obtained by the model stability analysis, the optimal feedback gain can be obtained by the improved Pareto sorting algorithm and Vague-sets decision method under the multi-objective optimal model which is at the basis of ensuring the stability, control precision of the system, then effective inner loop control with time-delay compensation and optimization will be achieved. Finally, the corresponding experimental platform will be constructed to verify the validity of the proposed method, and the project will provide important guidance for the design of this type system.

针对时滞对基于图像视觉伺服机械臂系统控制精度产生影响展开内外环控制时滞补偿与优化研究。对外环，提出目标运动中噪声相关性消除方法，构建Kalman有色噪声滤波器解决目标图像时滞补偿问题，通过优化系统采样及执行时间优化Kalman滤波步数，结合目标运动实际情况对目标图像进行时滞补偿估计及融合，提升目标图像及图像雅可比矩阵估计精度进而提升外环系统控制精度；对内环，构建机械臂时滞反馈动力学模型，针对其无限维、非线性、部分关节周期运动的特征，将其线性化结合低通滤波时间连续近似法转换为有限维不显含时滞反馈动力学模型，建立其前后两个离散时间状态间映射矩阵，寻求映射矩阵同反馈增益的关联，由系统稳定性获得反馈增益空间，提出改进Pareto排序和Vague集决策法找到兼顾系统稳定性和控制精度的多目标优化反馈增益，实现系统内环控制时滞补偿及优化。最后构建相应实验平台验证所提方法有效性，对该类系统设计提供重要指导

基于图像的视觉伺服机械臂控制基本未考虑系统时延带来的影响。图像获取传输处理产生时延，使当前时刻的目标图像及图像雅可比矩阵产生误差，该误差传递给内环伺服控制机构，影响机械臂控制精度。 . 课题展开该类系统控制时延补偿及优化研究。对外环，为提升目标图像及图像雅可比矩阵估计精度，提出常规模式下构建时延有色噪声情况下Kalman滤波器并结合时延来源进行滤波器补偿步数的确定完成目标图像及图像雅可比矩阵补偿估计，执行中增加眼在手及主动摄像机系统模式下的研究，图像补偿估计由当前时刻图像及延迟期间图像增量方式获得，包括机械臂末端执行器和目标图像变化因素分析及映射关系的确立、图像处理时间序列分析及延迟时间获取、图像补偿估计、图像雅可比矩阵更新，更全面地考虑系统工况。对内环，针对机械臂系统时延、非线性、部分关节呈现周期、混沌运动的特征，研究时延系统稳定性、动力学分析与控制问题，提出机械臂不显含时延反馈的动力学系统模型，结合系统稳定性及动态性能的目标，进行多目标优化算法研究，从而获得更好的控制效果。. 研究中的一些结果如下：(1)由视觉伺服处理周期获取及视觉采样周期独立优化调整策略进行补偿步数的确定，展开滤波补偿验证的研究，研究结果证明所提方法有效性。(2)除构建时延有色噪声情况下Kalman滤波器，提出了具有量测时延和有色量测噪声的非线性系统高斯滤波器，一般形式高斯滤波框架可由本研究退化所得。(3)增加眼在手、主动摄像机系统下的目标图像、图像雅可比矩阵时延补偿估计的内容，更全面地考虑系统工况。(4) 通过研究时延系统稳定性、动力学分析与控制问题，为机械臂这类时延、非线性、关节呈现周期混沌运动的系统提供稳定性及控制的研究支撑，同时实现机械臂不显含时延反馈控制动力学系统的转换，方便系统稳定性研究，简化了控制系统设计；通过多目标优化算法获得系统优化控制。 . 课题研究对提升机械臂控制精度、保障控制系统稳定性、简化控制系统，推广此类系统应用具有重要意义。