基于深度学习的机器人抓取虚拟仿真实验教学系统  

Virtual simulation experiment teaching system of robot grasping based on deep learning

作　　者：张飞[1] 尚伟伟[1] 何浩源 丛爽[1] 王大欣 ZHANG Fei;SHANG Weiwei;HE Haoyuan;CONG Shuang;WANG Daxin(School of Information Science and Technology,University of Science and Technology of China,Hefei 230026,China)

机构地区：[1]中国科学技术大学信息科学与技术学院,安徽合肥230026

出　　处：《实验技术与管理》2022年第1期173-177,共5页Experimental Technology and Management

基　　金：安徽省教育厅虚拟仿真实验教学项目(2020xfxm82);中国科学技术大学本科生质量工程项目(2021xjyxm012);国家自然科学基金项目(51675501)。

摘　　要：该文以基于深度学习的机器人抓取研究成果为基础,利用虚拟仿真技术,开发了机器人抓取实验教学系统。该系统形成了关键基础知识的模块化学习、数据集的创建与扩增、卷积神经网络的设计与训练、工业与生活服务场景的搭建,以及机器人抓取仿真实验验证等层次递进的实验过程,融合了理论知识考核、实验实践测评以及实验报告分析等在线评价方法,降低了实验教学成本,提高了教学质量和学习效率。Based on the research results of robot grasping with deep learning,this paper develops a robot grasping experimental teaching system by using virtual simulation technology.The system forms a progressive experimental process,including modular learning of fundamental basic knowledge,creation and expansion of data sets,designing and training of convolutional neural networks,construction of industrial and life service scenarios,and verification of robot grasping simulation experiments.Furthermore,the system integrates online evaluation methods such as theoretical knowledge assessment,experimental practice assessment,and experimental report analysis,which reduces practical teaching costs and improves teaching quality and learning efficiency.

关 键 词：深度学习 机器人抓取 虚拟仿真 实验教学 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置] G642.0[自动化与计算机技术—控制科学与工程]