Vision-based Tactile Sensor using Random Pattern

Abstract

Tactile sensors are being used in various fields such as automated factories and human collaboration, and have been developed in various technological ways. Most contact discrimination methods are performed through sensing of the physical aspect. On the other hand, in recent years, a vision-based tactile sensor that can obtain more information by replacing the existing method with only visual data of an image using a camera sensor has been proposed. Existing vision-based tactile sensors have markers as a standard to detect changes or have a pattern of a certain shape and arrangement. We propose a vision-based tactile sensor that can estimate the position of a contact with an easy design and compact hardware size. As a vision-based tactile sensor is a sensor that acquires various information about contact through image processing of visual data, it is possible to update the function of a tactile sensor by making good use of the fact that it is possible only through algorithm development without a separate hardware reconfiguration. We plan to continue our research to obtain more detailed contact information such as accuracy, force, and direction of force.