Enhanced braille recognition based on piezoresistive and piezoelectric dual-mode tactile sensors

Abstract

Human Tactile perception can be divided into pressure perception and slip perception according to the mode of contact. At present, however, tactile sensors can not have the same good performance as human pressure and slip perception. In this paper, a piezoresistive and piezoelectric dual-mode tactile sensing scheme is proposed based on human finger tactile skin structure. PVDF polymer is used as the sensitive material in piezoelectric layer, which has good dynamic response and high sensitivity. Graphene and carbon black are used as the sensitive material in piezoresistive layer, the array process is simple and high resolution can be achieved. The fast response of piezoelectric signal can be used to segment the piezoresistive signal and improve the recognition accuracy when the tactile sensor is used for slip sensing. At the same time, the piezoelectric layer can also reflect the velocity information and direction information. In recognition method, according to the feature that the continuous tactile signal has both temporal and spatial features in the process of sliding, CNN-LSTM is proposed in this paper. The tactile sensors is integrated into the robot arm to detect 20 kinds of braille. At the same speed and the same force, the recognition rate of bimodal braille recognition is 90.58%. In addition, the model can be well recognized at different speeds and sliding directions, and the recognition rate is 84.2%. Finally, the tactile sensor is wore on the index finger of human hand, the average recognition rate of touch braille is 77.5%, which shows that the algorithm has good robustness, and the speed of recognition is twice as fast as that of the normal blind. In the future, it can be applied to blind reading, surface detection, texture identification and other fields.