{"title":"Aerial-biped: a new physical expression by the biped robot using a quadrotor","authors":"Azumi Maekawa, Ryuma Niiyama, S. Yamanaka","doi":"10.1145/3214907.3214928","DOIUrl":null,"url":null,"abstract":"We present a biped robot which can move agiler than conventional robots. Our robot can generate bipedal walking motion automatically using the proposed method. By using a quadrotor for balance and movement it is possible to make an agiler movement, and generate a gait interactively and in real time according to the motion of the quadrotor using the optimized control policy of the legs. Our system takes the velocity of the quadrotor as an input and legs motions are produced so that the velocity of the foot in contact with the ground to zero, and bipedal walking motion is generated. The control policy is optimized using reinforcement learning with a physics engine.","PeriodicalId":370990,"journal":{"name":"ACM SIGGRAPH 2018 Emerging Technologies","volume":"77 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2018-08-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"1","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"ACM SIGGRAPH 2018 Emerging Technologies","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1145/3214907.3214928","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 1
Abstract
We present a biped robot which can move agiler than conventional robots. Our robot can generate bipedal walking motion automatically using the proposed method. By using a quadrotor for balance and movement it is possible to make an agiler movement, and generate a gait interactively and in real time according to the motion of the quadrotor using the optimized control policy of the legs. Our system takes the velocity of the quadrotor as an input and legs motions are produced so that the velocity of the foot in contact with the ground to zero, and bipedal walking motion is generated. The control policy is optimized using reinforcement learning with a physics engine.