摘要
为了解决采摘机器人作业过程中果实振荡造成目标识别不准确的问题,提出了一种针对运动果实的帧间差分法的扰动识别方法,并在帧间差分方法中引入了分裂迭代和模糊控制算法,实现了帧间差分背景图像的分离和子图像的有效聚类。依据该识别方法,对采摘机器人的目标逼近方法进行了改进,从而得到了更加准确的目标空间位置获取方法。为了验证该方法对运动果实目标逼近的有效性,采用虚拟仿真和机器人样机试验相结合的方法,进行了运动果实空间坐标获取和果实采摘试验。结果表明:采用分裂迭代模糊聚类的帧间差分方法,可以有效地对运动目标进行识别,识别误差较低,获得的位置坐标较为准确,可以满足果实采摘机器人的设计需求。
In order to solve the precision problem of target recognition for fruit picking robot oscillation caused in the process,it put forward a motion fruit frame difference disturbance identification method,and the frame difference method is introduced in splitting iterative algorithm and fuzzy control algorithm,which realized the effective separation of clustering and sub image points the background image of the frame difference.According to the identification method of picking robot,the target approximation method was improved to obtain the target space position more accurately.In order to verify the method of moving fruit object approximation,the method combined the virtual simulation and robot prototype experiments to get the movement of space coordinate acquisition and fruit picking fruit experiment.The experimental results show that the splitting iterative fuzzy clustering frame difference method can effectively identify the moving target with low identification error and more accurate position coordinates,which can meet the design requirements of the fruit picking robot.
出处
《农机化研究》
北大核心
2018年第4期58-61,224,共5页
Journal of Agricultural Mechanization Research
基金
河南省自然科学基金项目(2015GZC155)
南阳市科技攻关项目(2014GG059)
关键词
采摘机器人
目标逼近
分裂迭代
模糊聚类
帧间差分
picking robot
target approximation
splitting iteration
fuzzy clustering
frame difference