Classification of Pinching Action in Children using a Tomographic Tactile Sensor

Fine motor skills with the hands and fingers of children are useful for developmental assessment. To enable the sensing of subtle actions of the hand and fingers of children in a natural situation, we developed a cylindrical sensing device with a tomographic tactile sensor based on the concept of an object-based sensing system. U sing this device, we conducted an experiment to classify the number of fingers pinched and the pinching direction in children. The classification accuracy was calculated for each of the reconstructed images and the measured voltage signals based on the data collected during the experiment. Consequently, a maximum classification accuracy of 65.5% was achieved, which was above chance level (16.6
^%
). Furthermore, the classification accuracy obtained using the measured voltage signal was demonstrated to be higher than that obtained using the reconstructed image. These accuracies were lower than those of adult participants in our previous study, which may have been influenced by the child's small hand and finger size and weak grip strength. Although there remains room for improvement, the results provide a basis for a new evaluation method for subtle actions of the hand and fingers, which will contribute to the development of novel methodologies for developmental research.