A Bio-Mimetic Fingertip that can Detect Force and Vibration Modalities

For humanoid robots which interact with objects in unstructured environments, tactile sensing is important. It helps the robot to evaluate surface properties of the objects it interacts with. Texture identification and discrimination is an ability of such kinds of robots. In this paper, the development of a bio-mimetic fingertip that has accelerometers and force sensors to detect micro-vibration and force modalities is introduced. Furthermore, the ability of the fingertip in discriminating materials using textures is investigated. An artificial intelligence based approach is proposed with four parameters as classifiers for the ANN. The classifiers are the mean, standard deviation, and energy of detailed and approximate values of discreet transformation of the convoluted signal of accelerometer sensor signals. The method shows promising results as it was able to discriminate seven materials, six fabrics and one metal surface with 55% accuracy.