Abstract
The purpose of this study was to develop an air actuated robot finger having a tactile sense similar to a human finger. The result was an artificial finger which simulated the sensitivity function system of a human by combining the appropriate sensors and a neural network system. A sensory evaluation of humans by the SD method and multivariable data factor analysis was performed in order to clarify the human tactile sensitivity mechanism. The results were used to decide the type of sensors needed and were used as teaching models for the neural network. After the trainings, the artificial finger reached a level that could distinguish various materials like as a human, and an evaluation of its precision was carried out using the generalized functional results of the neural network.
