Design and tactile classification of flexible tactile sensor for soft gripper

Abstract

Tactile object recognition is very important in robot sorting and handling system. The existing researches on tactile recognition mainly focus on rigid grippers, which cannot be directly applied to soft grippers with better safety and applicability. In this paper, a tactile sensor for soft gripper is proposed. The tactile sensor consists of a carbon composite Velostat and a flexible printed circuit designed for soft gripper. The sensor is equipped with signal scanning, amplifying and processing systems. A number of performance tests were carried out on the designed tactile sensor to prove that it has good performance in sensing sensitivity, loading stability, repeatability. In addition, tactile sensors are applied to tactile classification. The two fingers of the flexible gripper were loaded with a sensor and nine different objects were contacted. The tactile information of the objects was obtained through the stable gripper, and the tactile data was trained and classified by the support vector machine. Finally, the tactile recognition accuracy of the double-exponent data was 91.1%.