Kanta Watanabe, Akio Numakura, S. Nishide, M. Gouko, Chyon Hae Kim
{"title":"高效的身体抖动,提高了机器人的牵引运动","authors":"Kanta Watanabe, Akio Numakura, S. Nishide, M. Gouko, Chyon Hae Kim","doi":"10.1109/ICMA.2015.7237650","DOIUrl":null,"url":null,"abstract":"This paper discusses the learning through body babbling, which is the initial stage of development and learning of human, from the view point of constructive approach for the recognition and behavior architectures of human. In previous researches, the body babbling and learning in the developmental process of drawing manipulation is modeled with two processes, a random joint angle generation process and an offline learning process with neural network. However, there is much gap between these models and the development of human. Human's developmental process is featured by the two processes, a judgment process for exploratory behaviors and an online incremental learning process. In this research, we propose a babbling-and-learning model that includes a judgment process for planned exploratory behaviors and an online incremental learning process. The online incremental learning process is modeled by Continuous System (Dynamics) Learning Tree (CSLT, DLT). CSLT realizes similar learning with neural network with an additional feature, online incremental learning. The proposed model introduces ε-greedy method for this judgment. After joint angles are executed, CSLT learns the obtained data by its online incremental learning process. As results of the validation of the proposed model, the proposed model gathered more number of effective data. The learning model decreased its prediction error faster than the previous model.","PeriodicalId":286366,"journal":{"name":"2015 IEEE International Conference on Mechatronics and Automation (ICMA)","volume":"54 4 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2015-09-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"5","resultStr":"{\"title\":\"Efficient body babbling for robot's drawing motion\",\"authors\":\"Kanta Watanabe, Akio Numakura, S. Nishide, M. Gouko, Chyon Hae Kim\",\"doi\":\"10.1109/ICMA.2015.7237650\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"This paper discusses the learning through body babbling, which is the initial stage of development and learning of human, from the view point of constructive approach for the recognition and behavior architectures of human. In previous researches, the body babbling and learning in the developmental process of drawing manipulation is modeled with two processes, a random joint angle generation process and an offline learning process with neural network. However, there is much gap between these models and the development of human. Human's developmental process is featured by the two processes, a judgment process for exploratory behaviors and an online incremental learning process. In this research, we propose a babbling-and-learning model that includes a judgment process for planned exploratory behaviors and an online incremental learning process. The online incremental learning process is modeled by Continuous System (Dynamics) Learning Tree (CSLT, DLT). CSLT realizes similar learning with neural network with an additional feature, online incremental learning. The proposed model introduces ε-greedy method for this judgment. After joint angles are executed, CSLT learns the obtained data by its online incremental learning process. As results of the validation of the proposed model, the proposed model gathered more number of effective data. The learning model decreased its prediction error faster than the previous model.\",\"PeriodicalId\":286366,\"journal\":{\"name\":\"2015 IEEE International Conference on Mechatronics and Automation (ICMA)\",\"volume\":\"54 4 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2015-09-03\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"5\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2015 IEEE International Conference on Mechatronics and Automation (ICMA)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/ICMA.2015.7237650\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2015 IEEE International Conference on Mechatronics and Automation (ICMA)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICMA.2015.7237650","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Efficient body babbling for robot's drawing motion
This paper discusses the learning through body babbling, which is the initial stage of development and learning of human, from the view point of constructive approach for the recognition and behavior architectures of human. In previous researches, the body babbling and learning in the developmental process of drawing manipulation is modeled with two processes, a random joint angle generation process and an offline learning process with neural network. However, there is much gap between these models and the development of human. Human's developmental process is featured by the two processes, a judgment process for exploratory behaviors and an online incremental learning process. In this research, we propose a babbling-and-learning model that includes a judgment process for planned exploratory behaviors and an online incremental learning process. The online incremental learning process is modeled by Continuous System (Dynamics) Learning Tree (CSLT, DLT). CSLT realizes similar learning with neural network with an additional feature, online incremental learning. The proposed model introduces ε-greedy method for this judgment. After joint angles are executed, CSLT learns the obtained data by its online incremental learning process. As results of the validation of the proposed model, the proposed model gathered more number of effective data. The learning model decreased its prediction error faster than the previous model.