In this paper, we propose a syntactic pattern recognition method for non-schematic drawings, based on a new attributed graph grammar with flexible embedding. In our graph grammar, the embedding rule permits the nodes ...
详细信息
A novel run-time algorithm is devised to eliminate dynamically unnecessary connections in neural networks that can be trained through supervised *** the performance error is being minimized,redundancy of neural connec...
详细信息
A novel run-time algorithm is devised to eliminate dynamically unnecessary connections in neural networks that can be trained through supervised *** the performance error is being minimized,redundancy of neural connections can be removed by applying a mutual suppression rule over the relevant set（s） of weight *** the best connections will finally overcome the struggle against the inhibitory ***,if a neuronal unit receives signals from units #1,…,#n,then the redundancy-reducing correction forω,which weights the connection from #j,is given by-εsgn（ω）(ω～2 +…+ω +ω+…+ω_n).This is a lateral inhibition acting on the parameters,not on the *** algorithm has been incorporated into a neural network of the standard backpropagation type. Some application examples are given to demonstrate its usefulness,including the problem of finding the minimum number of hidden units in certain logic tasks.
The authors discuss active tactile sensing for detecting a contact point between a multifingered robot hand and an unknown object. They define the concept of actively sensible meaning that an active sensing motion can...
详细信息
The authors discuss active tactile sensing for detecting a contact point between a multifingered robot hand and an unknown object. They define the concept of actively sensible meaning that an active sensing motion can be planned for a sensing event. Assuming more than two sensing events, conditional-minimum-external-sensor-realization is defined, where it is shown that it is possible to reduce the number of external sensors with a well-planned sensing algorithm. Based on these definitions, an active sensing algorithm for detecting the contact point with minimum-external-sensor-realization is proposed. This approach remedies several disadvantages found in conventional approaches. The approach was tested through experiments using a single-finger model with a fingertip tactile sensor.< >
A state estimation method for an energy stochastic system with a decibel observation mechanism is presented. The problem is to get a decibel-valued estimate of an energy state variable by using the decibel-valued nois...
详细信息
A state estimation method for an energy stochastic system with a decibel observation mechanism is presented. The problem is to get a decibel-valued estimate of an energy state variable by using the decibel-valued noisy observation data, where the stochastic system of concern is physically driven on energy scale. The main attention is paid to matching between the physical energy principle and the decibel observation mechanism on the human evaluation side. The present estimation method is based on an expansion form of the Baye's theorem, and any kinds of statistics can be estimated. The effectiveness of the theory has been confirmed by application to room acoustics data.< >
The authors discuss the self-posture changeability (SPC) for a general 3-D link system. The series of motions brought about by SPC is called self-posture changing motions (SPCM). SPC and SPCM are considered assuming t...
详细信息
The authors discuss the self-posture changeability (SPC) for a general 3-D link system. The series of motions brought about by SPC is called self-posture changing motions (SPCM). SPC and SPCM are considered assuming that the link system moves slowly enough to suppress any dynamic effects. It is shown that, according to a particular combination of link posture, joint assignment, and object shape, there exist a deadlock state, where no further angular displacement can be imparted to the position-controlled joint. It is also shown that there are two different types of deadlock states which never result in any SPCM. The authors introduce (w, omega )-posture changeability, the characteristic with which a link can change posture for every possible force within a cone of a prime axis w and a solid angle, omega . The necessary and sufficient conditions leading to (w, omega )-posture changeability are shown, for a given contact point and frictional conditions. a sufficient condition is included required to realize SPCM under the assumption that the contact point moves continuously over the object and the link-object system never results in a deadlock state.< >
The discrimination of gas-liquid two-phase flow patterns has so far been made mainly by visual observation; and various flow parameters characterizing the flow, such as fluctuating pressures, void fraction etc., have ...
详细信息
The discrimination of gas-liquid two-phase flow patterns has so far been made mainly by visual observation; and various flow parameters characterizing the flow, such as fluctuating pressures, void fraction etc., have been employed as quantitative tools for this purpose. It would be useful for the prediction and prevention of accidents experienced in thermal engineering plants dealing with the flow if it could be found the effective techniques for analyzing void fraction signals detected for diagnosing plant operation states. This paper presents pattern recognition and clustering techniques for analyzing the cross-sectional mean void fraction signals in gas-liquid two-phase flow based on an AR-model with bias components. The pattern recognition method based on the Bayes decision rule was applied for the recognition of flow regimes which were determined by visual observations. The clustering methods based on K-mean algorithm and mixture probability algorithm were applied for the determination of the new boundaries on a superficial gas-liquid velocities diagram. The effectiveness of the discussed techniques were confirmed.
This paper deals with the fuzzy drive control of an autonomous mobile robot. First, a fuzzy model (fuzzy expert system) for the human thought process is designed by using the idea of fuzziness, and secondly, it is app...
详细信息
This paper deals with the fuzzy drive control of an autonomous mobile robot. First, a fuzzy model (fuzzy expert system) for the human thought process is designed by using the idea of fuzziness, and secondly, it is applied to the autonomous mobile robot. The autonomous mobile robot is being investigated very actively in many research laboratories of companies and universities around the world. For example, there are the unmanned automated carrier in a factory, the working robot in extreme situations, and the house service robot [1, 3]. A human makes a proper decision and puts the decision into a proper action instantly under fuzzy and insufficient information. The fuzzy drive control system architecture is a hierarchy that is constructed by a visual sensor subsystem, a motor drive unit, and a fuzzy drive expert system which manages these subsystems. The visual sensor subsystem which perceives the outer environment such as the road, the obstacles, etc. is comprised of a TV camera and an image processing unit. The motor drive unit (motor controller) has the function of controlling the speed and the steering of the mobile robot by changing the revolution rate of motors mounted on the robot. The fuzzy drive expert system which is the core of the fuzzy drive control system recognizes the environment with the image data and sends control commands to the subsystems. By these commands from the drive expert system, each subsystem motivates the self-subsisting control function to achieve that purpose. This expert system has an inference unit and rules, such as recognition rules and control rules, which were obtained experimentally. As an application of the fuzzy drive control, experiments of driving an autonomous mobile robot on a straight road and one with corners are performed. From the experimental results, it was found that the robot control system has the faculty of flexibile circumstantial decision about equal to what a human is able to realize.
Static characteristics of a levitational mechanism are studied. The levitational mechanism consists of a high-T(c) superconductor tile (type-II superconductor) and a magnet arrangement of the same size bar magnets wit...
Static characteristics of a levitational mechanism are studied. The levitational mechanism consists of a high-T(c) superconductor tile (type-II superconductor) and a magnet arrangement of the same size bar magnets with alternating magnetic pole pattern. The levitation pressures have hysteresis loops because of the flux pinning effect. By using the alternating pole pattern of magnets larger static levitation pressure in proportion to the arrangement can be obtained over a wide area. Moreover the levitation pressure can be optimized with respect to the width of a bar magnet of the alternating pole pattern.
Drag pressures of a levitation mechanism are investigated. The levitation mechanism consists of a high-T(c) superconductor tile (type II superconductor) and a set of alternating-polarity bar magnets of the same size. ...
Drag pressures of a levitation mechanism are investigated. The levitation mechanism consists of a high-T(c) superconductor tile (type II superconductor) and a set of alternating-polarity bar magnets of the same size. Relationships of the drag pressure to the distance between the magnets and the superconductor tile and to the width of a bar magnet are investigated. The drag pressure becomes small rapidly with increasing distance. The drag pressure becomes large with the superconductor-magnets distance decreasing. If the distance is fixed there exist a width of a bar magnet providing a maximum drag pressure. The drag pressure is also measured when the superconductor is moved in the direction of the alternating polarity.
暂无评论