摘要
针对机器人在抓取电子元器件中目标实时精准定位的问题,应用点云配准技术,提出了一种扩展的二进制特征描述子(EB_SHOT)的电子元器件点云配准算法。首先,对采集的电子元器件点云进行ISS关键点提取,获取点云表面均匀分布且具有显著特征的点集。其次,对关键点采用SHOT特征描述和位置空间信息相结合的方法生成E_SHOT特征描述子,并将其编码为EB_SHOT特征描述子。最后,基于FLANN特征点搜索路径进行SAC-IA和ICP相结合的点云配准方法,得到点云之间的空间坐标变换矩阵,从而获得电子元器件的位姿信息。实验结果表明,使用这里方法进行电子元器件点云配准,在配准精度和速度上均有所提高,满足机器人抓取的要求。
Aiming at the problem of real-time and accurate positioning of targets in the robot grasping electronic components,using point cloud registration technology,an extended binary feature descriptor(EB_SHOT)electronic component point cloud registration algorithm is proposed.First,perform ISS key point extraction on the collected electronic component point cloud to obtain a point set that is uniformly distributed on the surface of the point cloud and has significant characteristics.Secondly,for the key points,the method of combining SHOT feature description and location spatial information is used to generate E_SHOT feature descriptor,and encode it as EB_SHOT feature descriptor.Finally,based on the FLANN feature point search path,the point cloud registration method combining SAC-IA and ICP is performed to obtain the spatial coordinate transformation matrix between the point clouds,so as to obtain the pose information of the electronic components.The experimental results show that the use of this method for electronic component point cloud registration improves the registration accuracy and speed,and meets the requirements of robot grasping.
作者
孙建荣
许俏
顾寄南
季琳琳
SUN Jian-rong;XU Qiao;GU Ji-nan;JI Lin-lin(School of Mechanical Engineering,Jiangsu University,Jiangsu Zhenjiang 212000,China)
出处
《机械设计与制造》
北大核心
2023年第6期53-56,共4页
Machinery Design & Manufacture
基金
国家自然科学基金(51875266)。
关键词
机器视觉
点云配准
关键点提取
特征描述
位姿估计
Machine Vision
Point Cloud Registration
Key Point Extraction
Feature Description
Pose Estimation