Gripping status classification using an array of micro cantilever type tactile sensor

We have designed and fabricated a tactile sensors array with three inclined micro-cantilevers on a single sensor embedded in elastomer. Each sensor element can detect both normal and shear stress. In this paper, we propose classification method of gripping status using sensor array and a machine learning algorithm. By integrating multiple sensor data using machine learning algorithm, our method allows to detect partial slipping on contact surface as an initial sign of slipping. Through experiments, we confirmed that our method achieves better classification ratio than existing methods using single sensor data.