摘要
为提高机器人定位精度,解决工业机器人标定环节中参数辨识精度低,冗余参数无法辨识等问题,提出一种分步式自适应学习参数辨识方法。该方法采用等步长区域搜索,步长设置过小,区域搜索容易陷入局部最优而终止辨识,步长设置过大,单步搜索分辨率下降,辨识精度降低。为优化该算法,提出二分法区域搜索方法。运动模型上,为克服D-H模型的奇异性问题,采用D-H模型与MDH模型相结合的方法。辨识环节,首先遍历合适的测量位姿,然后补偿独立参数,接着采用区域搜索的方法对剩余参数进行误差辨识。最后运用MATLAB软件仿真,末端位置误差的最大偏差降低至0.2mm以内,证明该算法改进后辨识精度更高,辨识结果更加稳定。
In order to improve the positioning accuracy of the robot and solve the problems of low accuracy of parameter identification and unrecognizable redundant parameters in the calibration process of industrial robots,a step-by-step adaptive learning parameter identification method is proposed.The method uses an equal step area search,the step size is set too small,the area search is easy to fall into the local optimum and the identification is terminated,the step size is set too large,the single step search resolution is decreased,and the identification precision is lowered.In order to optimize the algorithm,a dichotomy region search method is proposed.In the motion model,in order to overcome the singularity problem of the D-H model,a combination of the D-H model and the MDH model is adopted.In the identification process,the appropriate measurement pose is traversed first,then the independent parameters are compensated,and then the residual parameters are identified by the region search method.Finally,using MATLAB software simulation,the maximum deviation of the end position error is reduced to less than 0.2mm,which proves that the improved algorithm has higher identification accuracy and the identification result is more stable.
作者
高兴宇
付树兵
李煜
陆佳琪
GAO Xing-yu;FU Shu-bing;LI Yu;LU Jia-qi(College of Mechanictil and Electrical Engineering,Guilin University of Electronic Technology,Guangxi Guilin 541004,China)
出处
《机械设计与制造》
北大核心
2022年第10期285-289,294,共6页
Machinery Design & Manufacture
基金
广西创新驱动发展专项资金项目(桂科AA18118002-3)
桂林电子科技大学研究生教育创新计划资(2018YJCX05)
桂林电子科技大学研究生教育创新计划资助项目(2019YCXS006)。
关键词
工业机器人
参数辨识
分步式
区域搜索
二分法
Industrial Robot
Parameter Identification
Step-by-Step
Area Search
Dichotomy
