不规则金属物料的抓取位姿实时检测方法研究

Study on Real-time Detection of Grab Pose of Irregular Scrap Metal

为解决机器人抓取不规则金属物料时存在的抓取位姿难以准确判断的问题,在图像处理算法的基础上,提出一种基于轮廓和形心特征的抓取位姿判断方法。该方法将金属物料分为两种类型,一种是形心落在轮廓范围以外,另一种是形心落在轮廓范围以内。对于第一种物料,采用直线求交法确定抓取位姿,对于另外一种提出射线求交法确定抓取位姿。两种方法均是以过形心的直线或射线与物料轮廓相交点的最短欧式距离作为主要判断依据,确定最优的抓取位置。实际抓取实验表明:该算法可以实现对不规则金属物料的实时、稳定、准确抓取。

To challenge the difficulty in judging grasp position and pose with accuracy in operation of irregular metal material trapping by robot,by image processing algorithm,a method of judging grasp position and pose is proposed based on contour and centroid features.By the method,metal materials are divided into two types:one centroid falls outside the contour range,and the other within the contour range.For the first one,the straight line intersection method is adopted to determine the grasping position and pose,while for the other,the ray intersection method is proposed to determine the grasping position and pose.Both of the two methods take the shortest Euclidean distance of the intersection point between the line or ray through the centroid and the material profile as the main judgment basis to determine the optimal grasping position.The actual grasping experiment shows that the algorithm can realize real-time,stable and accurate grasping of irregular metal materials.