{"title":"基于web的基于深度神经网络的机器人物联网三维可视化分布式路径规划分类","authors":"Z. Iklima, T. M. Kadarina","doi":"10.30880/jst.2021.13.02.006","DOIUrl":null,"url":null,"abstract":"Internet of Robotic Things (IoRT) distributes heterogeneous intelligences among devices and platforms. A distributed control of a three-degree-of-freedom (3-DOF) robot manipulator is integrated with web-based 3D visualization. An asynchronous protocol was utilized to broadcast kinematic data of a 3-DOF robot manipulator between platforms. However, kinematic data computed using inverse kinematic equations directly cannot identify the singularity issue of robot manipulator. Singularity avoidance required to prevent robot component or joint from damage. Therefore, this study proposed a deep neural network approach as a classification-based of manipulator robot path planning to avoid singularity issues. Deep neural network (DNN) was trained in 12 minutes, 52 seconds in 500 iterations. Training accuracy measured with value 96,23 percent, validation accuracy measured with value 96,13 percent, and testing accuracy measured with value 96,48 percent Additionally, 3 DOF manipulator robot web-based 3D visualization was made using Web Graphics Library (WebGL). The distributed platform was tested successfully and can distribute and classify 2352 motions per second.","PeriodicalId":21913,"journal":{"name":"Songklanakarin Journal of Science and Technology","volume":"6 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2021-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":"{\"title\":\"Distributed Path Planning Classification with Web-based 3D Visualization using Deep Neural Network for Internet of Robotic Things\",\"authors\":\"Z. Iklima, T. M. Kadarina\",\"doi\":\"10.30880/jst.2021.13.02.006\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Internet of Robotic Things (IoRT) distributes heterogeneous intelligences among devices and platforms. A distributed control of a three-degree-of-freedom (3-DOF) robot manipulator is integrated with web-based 3D visualization. An asynchronous protocol was utilized to broadcast kinematic data of a 3-DOF robot manipulator between platforms. However, kinematic data computed using inverse kinematic equations directly cannot identify the singularity issue of robot manipulator. Singularity avoidance required to prevent robot component or joint from damage. Therefore, this study proposed a deep neural network approach as a classification-based of manipulator robot path planning to avoid singularity issues. Deep neural network (DNN) was trained in 12 minutes, 52 seconds in 500 iterations. Training accuracy measured with value 96,23 percent, validation accuracy measured with value 96,13 percent, and testing accuracy measured with value 96,48 percent Additionally, 3 DOF manipulator robot web-based 3D visualization was made using Web Graphics Library (WebGL). The distributed platform was tested successfully and can distribute and classify 2352 motions per second.\",\"PeriodicalId\":21913,\"journal\":{\"name\":\"Songklanakarin Journal of Science and Technology\",\"volume\":\"6 1\",\"pages\":\"\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2021-12-02\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"1\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Songklanakarin Journal of Science and Technology\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.30880/jst.2021.13.02.006\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"Multidisciplinary\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Songklanakarin Journal of Science and Technology","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.30880/jst.2021.13.02.006","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"Multidisciplinary","Score":null,"Total":0}
Distributed Path Planning Classification with Web-based 3D Visualization using Deep Neural Network for Internet of Robotic Things
Internet of Robotic Things (IoRT) distributes heterogeneous intelligences among devices and platforms. A distributed control of a three-degree-of-freedom (3-DOF) robot manipulator is integrated with web-based 3D visualization. An asynchronous protocol was utilized to broadcast kinematic data of a 3-DOF robot manipulator between platforms. However, kinematic data computed using inverse kinematic equations directly cannot identify the singularity issue of robot manipulator. Singularity avoidance required to prevent robot component or joint from damage. Therefore, this study proposed a deep neural network approach as a classification-based of manipulator robot path planning to avoid singularity issues. Deep neural network (DNN) was trained in 12 minutes, 52 seconds in 500 iterations. Training accuracy measured with value 96,23 percent, validation accuracy measured with value 96,13 percent, and testing accuracy measured with value 96,48 percent Additionally, 3 DOF manipulator robot web-based 3D visualization was made using Web Graphics Library (WebGL). The distributed platform was tested successfully and can distribute and classify 2352 motions per second.
期刊介绍:
Songklanakarin Journal of Science and Technology (SJST) aims to provide an interdisciplinary platform for the dissemination of current knowledge and advances in science and technology. Areas covered include Agricultural and Biological Sciences, Biotechnology and Agro-Industry, Chemistry and Pharmaceutical Sciences, Engineering and Industrial Research, Environmental and Natural Resources, and Physical Sciences and Mathematics. Songklanakarin Journal of Science and Technology publishes original research work, either as full length articles or as short communications, technical articles, and review articles.