(A) force enhanced soft palm gripper for object identification

Abstract

Soft gripper made of flexible materials are easily deformed and useful when holding various objects without complicated control. However, due to the deformability of the soft gripper, the gripping force is limited compared with the conventional rigid gripper. Also, in order to expand the application field of the soft gripper, it is required to have cognitive ability to get the information of the gripped object. But, the cognitive of soft grippers in previous researches only identified the objects with the different shape. In this study, we propose a gripper based on soft palm structure to solve these problems. We used the soft palm gripper to increase the gripping force of the soft gripper. In addition, by learning the fully-connected neural network based object classifier using the pneumatic data of the soft gripper, it is possible to identify objects of different shape and mass. Using our proposed soft gripper and classification model, it is possible to improve the strength and cognitive abilities of the soft gripper.