Designers of mechanical products frequently refer to engineering drawings which are stored as image data in databases to design a new mechanical product efficiently. Multiple mechanical parts are usually drawn on each engineering drawing. Therefore designers want to find engineering drawings containing parts similar to a query image in the shape of a part drawn on an engineering drawing. In this paper, we propose a novel similarity based partial image retrieval system for engineering drawings. A unique aspect of this system is that a graph representation is utilized to robustly find engineering drawings containing similar parts which are invariant to the size, position, and rotation. We verified the performance for the similarity based partial image retrieval system through experiments using industrial engineering drawings. The results show that the top five similar engineering drawings for every query image are always accurately retrieved by our proposed system. This finding suggests that this system could be useful for the reuse of stored engineering drawings.

This paper describes our multimedia information retrieval system, MIRACLES, for retrieving multimedia content such as web pages, business documents, and movies. MIRACLES creates information units, which are collections of related media items, such as an image and text explaining the image. MIRACLES arranges and displays the information units in 3-D space so that similar information units are located near each other. The user can narrow down the retrieval of the multimedia content by rearranging the information units in the display space. MIRACLES is already in practical use, and is employed in some commercial products and services. We are presently evaluating the use of MIRACLES by actual users.

Several position identification methods have been used for mobile robots. We propose a new method called "cooperative positioning system (CPS)". For CPS, we divide the robots into two groups, A and B. The group A remains stationary and acts as a landmark while group B is moving. Group B then stops and acts as a landmark for group A. This "dance" is repeated until the target position is reached. By using the concept of "portable landmarks", CPS has a far lower accumulation of positioning error than dead reckoning, and can work in three-dimensions which is not possible with dead reckoning; it can therefore work in uncharted environments. In this paper we outline the second prototype CPS machine model (CPS-II) and report the results of position identification experiments. Experimental results using this model give a positioning accuracy of o.4% for position and 1.0 degree for attitude.

A number of positioning identification techniques have been used for mobile robots. Dead reckoning is a popular method, but is not reliable when a robot travels long distances or over an uneven surface because of variations in wheel diameter and wheel slippage. The landmark method, which estimates the current position relative to landmarks, cannot be used in an uncharted environment. The authors propose a new method called "cooperative positioning with multiple robots". For cooperative positioning, the authors divide the robots into two groups, A and B. One group, say A, remains stationary and acts as a landmark while group B moves. The moving group B then stops and acts as a landmark for group A. This "dance" is repeated until the target robot position are reached. Cooperative positioning has a far lower accumulated positioning error than dead reckoning, and can work in three-dimensions which is not possible with dead reckoning. Also, this method has inherent landmarks and therefore works in uncharted environments. This paper discusses the positioning accuracy of the authors' method with error variances for an example with three mobile robots.>

An inverted pendulum presents a typical problem of controlling a nonlinear multivariable system. In the paper a discussion is presented of inverted pendulum control by a neural network. The neutral network learns a control law through the trial and error method. In the simulation and actual system, it was possible to invert the pendulum stably at the target position after some trial and error.>

A mobile robot with behavior controlled by a neural network, and its learning method are presented. The robot has four wheels and travels with two motors; it has twelve sensors for detecting internal conditions and changes in environment. The sensor signals are fed into the input layer of the network, and the network outputs motor control commands. The network model is divided into two subnetworks connected to each other with a short-term memory to process time-dependent data. The robot can learn various habits by changing the patterns to be taught. One example, the habits to play a cops-and-robbers game, was taught. Through training, the robots learned habits such as capture and escape.>