Suction cup array working also as tactile sensor to detect cups deformation using kcf and cnn

We propose a micro-suction cup array device with tactile sensing capabilities. The suction cup array device is equipped with a camera for capturing images of the contact surface. The camera reads the deformation of the suction cups. Image processing estimates the displacement of each suction cup contact surface, its direction, and the contact force distribution. A preliminary experimental device has been fabricated. The master shape of the micro cup was fabricated by 3D laser lithography and UV nanoimprinting processes. A hole in the suction cup was opened by the femtosecond laser beam. Kernelized Correlation Filter was applied to the acquired images and the movement of the contact surface of each suction cup could be estimated. Convolutional Neural Networks (CNN) was used to estimate the pressing force. The correctness (accuracy) of the produced CNN model was 87%, which is a practical accuracy.