Open AccessPouring from Deformable Containers Based on Tactile Information Using a Dual-Arm Robot
Author(s) -
Yuto Tsuchiya
Publication year - 2019
Publication title -
international journal of robotic computing
Language(s) - English
Resource type - Journals
ISSN - 2641-9521
DOI - 10.35708/rc1868-126254
Subject(s) - container (type theory) , robot , computer science , deformation (meteorology) , dual (grammatical number) , motion (physics) , artificial intelligence , computer vision , mathematics , simulation , mechanical engineering , engineering , materials science , composite material , art , literature
In this paper, we consider household robots that pour various contents from deformable containers. Such pouring is often seen incooking and refilling. To achieve this kind of pouring, we reduce the deformation of the container during pouring and thus carefully design thegrasping strategy: the palm of one hand supports the deformable container from the bottom and the other hand pulls up the container fromthe top. We apply the proposed system to pouring four different kinds ofcontents: breakfast cereal, coffee beans, flour, and rice. The experimentverifies that the proposed system successfully pours the four contents.To evaluate the system quantitatively, we measure 1) the deformation ofthe container using a motion capture system and 2) the success rate ofpouring. We verify that the dual-arm pouring reduced the deformationby 66% compared to a single-arm motion and that the success rate isgreater than 90%.