Fuzzy

240

Fuzzy Robotics and Fuzzy Image Understanding K a o r u Hirota Department of Computational Intelligence and Systems Science Interdisciplinary Graduate School of Science and Engineering Tokyo Institute of Technology 4259 Nagatsuta-cho, Midori-ku, Yokohama 226, Japan e-mail hirot [email protected]

A b s t r a c t . Two topics, i.e., fuzzy robotics and fuzzy image understanding, are introduced in this order. The author's group developed various types of fuzzy robots from early 1980s. Several typical examples are shown in this talk. The first one realizes a grasping and putting operation in industrial line based on fuzzy if then rule based control and fuzzy image decision tree techniques. Then a flower arrangement robot is introduced. It realizes the human artistic skill in the robot task based on fuzzy rules. The third one is 2D Ping-Pong robot that demonstrates the realtime control with image process- ing using a simple 16-bit personal computer. The forth is con- cerned with an irregular moving object shooting based on fuzzy chaos prediction method. Finally a group of autonomous mobile robots is presented. It realizes various group movements such as line formation and triangular shaped movement, where fuzzy neuro techniques are introduced. There exist several advantages in these approaches to robotics. Fuzzy technologies made it possible to introduce humans' skill in a human friendly manner and to design the system easily. In most application examples presented here it took for a few weeks for a few graduate students to implement the fundamental part of the system including the knowl- edge acquisition process. The costeffectiveness of these technol- ogies is another noteworthy point. Even a very small personal computer realized the real time control or the real time image understanding for these robotics examples. Some of the techniques presented have been applied to Japanese robotics applications in real industrial lines. Then various image understanding techniques based on fuzzy technology developed by the authors group have been surveyed. First fuzzy clustering is applied to the remote sensing images. It is a modified version of the well known FCM. Then a shape recognition algorithm is presented for a robotics assembling line. It is a fuzzy discriminant tree method for a real time use. Finally a fuzzy dynamic image understanding system is presented. It can understand the dynamic images on general roads in Japan, where a fuzzy frame based knowledge representation and a special kind of fuzzy inference engine are introduced. Experimental results are shown by an OHP and a slide projector.