Object identification by tactile sensors with improved sensor resolution

In the case that tactile sensor matrices are placed at the fingers of a robotic gripper, it is possible to identify objects during grasping. This feature may be useful e.g. for sorting workpieces. In this article we present an approach for object identification using tactile information. For verification a test scenario is developed where the robot should sort workpieces which are based on Lego bricks. In comparison to images of camera systems commonly used in robotics, the resolution of force images provided by tactile sensors is much lower. For that reason it may be in some case useful to increase the resolution of tactile measurement to perform some tasks of object identification successfully. It can be achieved by grasping the object on different positions and merge the measured values of the tactile sensor cells (taxeis). This article additionally includes an algorithm to improve the resolution of tactile sensors. For the verification of the proposed approach a test scenario is worked out, where the orientation of a Lego brick should be detected. For this purpose a method for identification of the different sides is developed. AII algorithms are implemented with a real robot system and experimental results are presented.