A deformable fingertip sensor assists the manipulator in distinguishing the hardness of the object

In this work, we propose and fabricate a deformable and arrayed fingertip sensor to assist the manipulator in sensing the hardness of objects. The sensor consists of a semi-cylindrical Ecoflex and an array of electronic skin, which is attached to a robotic gripper for griping objects of different hardness. The electrical signals generated by the fingertip sensor vary depending on the hardness of the gripped object. After analyzing the signals, with a fixed gripping force, we found that the harder the object, the greater the average intensity of the signals collected by the sensor. This is consistent with the regularity shown by the finite element analysis method. The data collected by the sensors were used to train the deep neural network and classify ten different hardness objects, and the results showed that the recognition accuracy was as high as 91.6%.