Paper folding is one of the most difficult tasks for multi-fingered robot hands because paper is deformable and its stiffness distribution is nonuniform. In this study, we aim to achieve dexterous paper folding by extracting some dynamic primitives.
Each primitive uses visual and force information, a physical model of a paper sheet for analyzing its deformation, a machine learning method for predicting its future state. In this paper, we propose a strategy to achieve valley folds of a sheet of paper twice in a row.
In the second fold, a crease line of the first fold disturbs accuracy of the folding. We propose some new manipulation techniques to solve the problem. Finally we show demonstrations of the paper folding achieved with high success rate.