基于三维点云的多指机械手抓取应用研究  

Application research of multi-finger robot hand grasping based on three dimensional point cloud

作　　者：占宏[1] 黄杰彬 杨辰光[1] ZHAN Hong;HUANG Jiebin;YANG Chenguang(School of Automation Science and Engineering,South China University of Technology,Guangzhou 510641,China)

机构地区：[1]华南理工大学自动化科学与工程学院,广州510641

出　　处：《机械传动》2025年第3期141-145,共5页Journal of Mechanical Transmission

基　　金：广东省高等教育教学研究和改革项目(粤教高函〔2023〕4号);华南理工大学校级教研教改项目(C9249450);华南理工大学第十批探索性实验项目(C9238440)。

摘　　要：【目的】在工业制造领域,执行抓取动作是机器人完成复杂作业任务的基础,因此,实现对未知物体的精确感知和准确抓取至关重要。搭建机械手抓取试验平台,提出了一种基于多角度视觉信息的机械手抓取规划方案。【方法】首先,采用深度相机从多视角采集未知物体点云图像并进行三维重建,从而获取未知物体位置和形状信息;其次,采用高斯过程拟合未知物体隐式曲面,并基于吸引子运动的机械手动态抓取规划方案,实现对未知物体的特制化抓取。【结果】通过试验验证了所提出方案具有速度快、精度高、适应性强等特点,能够适应不同环境和任务需求。所提方案对于机器人相关应用研究具有重要意义。[Objective]In the field of industrial manufacturing,the grasping action is the basis for robots to complete complex tasks.Therefore,it is very important to realize accurate perception and grasping of unknown and irregularly shaped objects for robots.A robot grasping experimental platform was built,and a grasping planning method based on multi-angle visual information was proposed.[Methods]Firstly,a depth camera was used to collect the point cloud image of the unknown object from multiple views,and it was reconstructed in three dimensions to obtain the position and shape information of the unknown object.Then,the Gaussian process was utilized to fit the hidden surface of the unknown object,and the robot dynamic grasping planning scheme based on the attractor motion was used to realize the customized grasping.[Results]The proposed grasping planning scheme was verified by the experiments to have the characteristics of fast speed,high precision and strong adaptability,and can meet the requirements of different environments and tasks,which is of great significance for practical application research related to robotics.

关 键 词：机器人 未知物体抓取 点云重建 高斯过程 

分 类 号：TP242.2[自动化与计算机技术—检测技术与自动化装置]