An effective distributed multi-agent system using local communication is proposed in this paper, which employs a state-based approach and achieves a complex task without any centralized agent. In general, many researches on cooperative multi-agent system employ a centralized approach which uses a centralized agent as supervisor. However, it has drawbacks that it is weak for a failure of the centralized agent and the task management becomes difficult as the number of agents increases. We propose a distributed-type cooperative multi-agent system in which tasks of each agent are arranged effectively by using local information. In the proposed model, each agent determines effective actions by using a situation of a whole system estimated from a local communication between neighboring agents. The proposed system offers a low computing cost and effective cooperative actions. The performance of the proposed system is confirmed by computer simulations on surrounding task.

A multi-modal tactile sensing system for a human-like tactile representation is proposed. The proposed system consists of the multimodal tactile sensor and a texture recognition block employing fuzzy inference. The sensor inspired by structure of human's fingertip detects pressure, thermal conductivity and strain by using a pressure-sensitive conductive rubber, a PVDF film and a strain gauge, respectively. By introducing two active motions which are pressing and tracing on a surface of an object, hardness, warmth, friction and roughness of a material can be recognized. Furthermore subjective material-recognition mechanism can be implemented using fuzzy rules that describe directly a human-like sense of touch in word. We will show material-recognition experiments using several materials found in our daily life and discuss the performance of the system for subjective recognition.

This paper describes a practical several moving objects area extraction using a half-cosine function wavelet network. The network decomposes an image into weighted basis functions corresponding to spatial frequency. In this method, objects are distinguished by using a difference of object's size. First, moving objects area can be extracted from the weight image of the basis functions which constructs the reconstructed image by using frame difference. After that objects with different size are distinguished by using level-image difference. The proposed method enables measurement and area extraction of objects simultaneously. Moving objects area extraction from an image stream which includes several different-size objects is investigated. We show that the proposed basis function based method has robustness against background noise.

In this paper, we propose an FPGA-based collision warning system for advanced automobile driver assistance systems or autonomous moving robots. The system consists of three function blocks: coarse edge detection using a resistive-fuse network, moving-object detection inspired by neuronal propagation in the hippocampus, and danger evaluation and collision warning using fuzzy inference. The first two functions are implemented in FPGAs. The system can detect moving objects with a speed range of 3-192 km/h with a sampling period of 30 ms for an input image of 320 x 256 pixels, and can output a warning against dangerous regions in the input image.

In this paper, a new morphological associative memory (MAM) is presented. The proposed model is based on the morphological associative memory without a kernel image. Although the model is susceptibility to noise in input patterns, the model is useful for practical applications. We try to improve the problem by introducing a block splitting approach to the recalling process. Furthermore, the proposed method also has a merit the size of memory matrices is small. We confirm the effectiveness of the proposed model by autoassociation of alphabet patterns to compare traditional approaches in terms of the noise tolerance and the size of memory matrices.

In this paper, we describe the design, modeling, simulation and implementation of an entertainment robot named "Jumping Joe". Jumping Joe, an artistic and agile robot, can perform several rapid movements such as fast wake up, jumping and somersault. In order to realize acrobatic movements, four different actuators which can create the high-speed movements are developed. Jumping Joe with eight DOF, which are two DOF in each hip joint and one DOF in each knee joint and each foot, can also perform the routine movements of a legged robot such as standing, walking etc. First we describe the principle of the developed actuators. Second we investigate the feasibility of the special movements performed by our artistic robot through simulations using 3D dynamical analysis software. Next we show experimentally that a prototype of our artistic robot "Jumping Joe" can demonstrate all the above rapid movements.

Nine basic fuzzy logic circuits employing p-ch and n-ch current mirrors are presented, and the fuzzy information processing hardware system design at a low cost with only one kind of master slice (semicustom fuzzy logic IC) is described. The fuzzy logic circuits presented here will be indispensable for a "fuzzy computer" in the near future.