ファジィプロダクションルールによる三軸触覚センサ装着ロボットハンドを用いた触覚認識

概要

論文の詳細を見る
The tactile image resulting from touching depends on contact conditions arising between the tactile sensor and the object. Moreover, tactile recognition is usually concluded such fuzzy linguistic expressions as "This object is very hard" or "The object is rather soft," which are well adaptable to fuzzy theory. An inference system using fuzzy production rules can quantify the fuzziness involved in the tactile image data and employ the fuzzy linguistic expressions for tactile recognition. The present tactile recognition system has been implemented by utilizing fuzzy production rules and LISP functions for manipulating the fuzzy set. Two kinds of experiments of hardness and shape recognition are performed using a robot hand equipped with three-dimensional tactile sensors. One is calibration tests which adjust membership functions using seven standard specimens for hardness and five for shape, respectively. The other is verification tests using unknown specimens which are similar to standard ones. Since it is found that the present system is capable of recognizing these standard and unknown specimens, the inference method using fuzzy production rules is effective for tactile recognition.
一般社団法人日本機械学会の論文
1995-08-25