基于AUKF的机器人视觉伺服模型预测控制方法研究

打开文本图片集

关键词：视觉伺服；图像雅可比矩阵；自适应噪声估计；无迹卡尔曼滤波；内点法；模型预测控制中图分类号：TN911.7-34；TP242 文献标识码：A 文章编号：1004-373X（2026）09-0156-06

AUKF-basedrobotvisualservomodelpredictivecontrol

Jia Yang，Zhang Jianye，Wu Zizhao （School ofMechanical Engineering，TiangongUniversity，Tianjin3Oo382，China）

Abstract：InviewofthepoorimageJacobianmatrixestimationacuracyrobotsystemconstraints，andcamerafieldof-view constraints inrobot visual servo，thispaper proposesarobotvisual servocontrol methdcombining adaptiveunscented Kalman filtering（AUKF）andinteriorpointmodelpredictivecontrol（IP-MPC）.Firstly，thetraditionalUKFalgorithmisnotaccurate enoughfornoiseestimation，soSage-Husafilteringisintroducedtoestimatetheprocessnoisecovariancematrixonlineto improvetheonlineestimationacuracyofimageJacobianmatrix.Secondly，fortheconstraintsofrobotsystemandcamerafieldof-view，themodelpredictivecontrollrisdesignedbyheinteriorpointmethodandtheconstraintproblemistrasfodito theproblemofminimizingquadraticprogramming torealizethetrackingcontrolofrobotvisualservosystem.Theexperiments showthattheconvergencespeedoftheproposedmethodisimprovedby38.1%，itsaverageerrorofimagefeaturepointsis reduced by 37.4% ，and theend-effectorspeed fluctuationof therobotis decreased，which demonstratethat the proposed method hassignificantimprovementsinvisualservocontrolaccuracy，andsystemresponsespeedandstabilityincomparisonwiththe traditionalUKFalgorithm.

Keywords：visualservo;imageJacobianmatrix;daptivenoiseestimation;UKF;interiorpointmethod;modelpredictivecotrol

0 引言

机器人视觉伺服（VisualServo）是利用机器人末端相机进行图像反馈来引导机器人运动的一种控制方式，在工业机器人、智能装配、精密加工等领域发挥着重要的作用。（剩余8153字）