All issues
Volume 45, Issue 11
Displaying 1-11 of 11 articles from this issue
- |<
- <
- 1
- >
- >|
Special Issue on SSI 2008—System and Information Technology toward Emergence of Innovation—
-
[in Japanese], [in Japanese], [in Japanese]2009 Volume 45 Issue 11 Pages 545
Published: 2009
Released on J-STAGE: November 03, 2011
JOURNAL FREE ACCESSDownload PDF (134K)
Paper
-
Yukio HORIGUCHI, Ryuichi FUKUJU, Tetsuo SAWARAGI2009 Volume 45 Issue 11 Pages 546-554
Published: 2009
Released on J-STAGE: November 03, 2011
JOURNAL FREE ACCESSAutomated systems are often structured into modes for providing enhanced capabilities to perform complex tasks, but they also increase the need for and difficulty of maintaining mode awareness, which may induce mode errors and eventually lead to automation surprises. Salient feedback on the mode status is necessary for the user to maintain mode awareness, and it should be carefully designed not to interfere in the user's primary task to be achieved in collaboration with the systems. For the purpose of developing a practical theory for effective and robust supports for maintaining mode awareness, this study focuses on action-effect relations as an intuitive human-machine communication channel. An action-effect relation is defined as an association of a user action with some observable effects on the system behavior, and it is expected to provide additional valuable information on the internal state of the system with no focal visual attention required. From this perspective, one hypothesis is made that different modes with a same or similar configuration of action-effect relations should be easily mixed up with one another. This paper tests this hypothesis through a driving simulator experiment where subject drivers were examined on their awareness of the operating mode status of ACC (Adaptive Cruise Control) during driving with the mode-rich automated system. Another experiment examines the contribution of differentiated configuration of action-effect relations to enhancing mode awareness.View full abstractDownload PDF (1132K) -
Makoto ITOH2009 Volume 45 Issue 11 Pages 555-561
Published: 2009
Released on J-STAGE: November 03, 2011
JOURNAL FREE ACCESSAvoiding over-trust in machines is a vital issue in order to establish intelligent driver support systems. It is necessary to distinguish systems for workload reduction from systems for accident prevention/mitigation. This study focuses on over-trust in an Adaptive Cruise Control (ACC) system as a typical driving support system for workload reduction. By conducting an experiment, we obtained a case in which a driver trusted the ACC system too much. Concretely speaking, the driver just watched the ACC system crashing into a stopped car even though the ACC system was designed to ignore such stopped cars. This paper investigates possible contributing factors to the driver' s over-trust in the ACC system. The results suggest that emerging trust in the dimension of performance may cause over-trust in the dimension of method or purpose.View full abstractDownload PDF (1031K) -
Eiji KONAKA, Tatsuya SUZUKI, Kazuya ASANO, Yoshitsugu IIJIMA2009 Volume 45 Issue 11 Pages 562-569
Published: 2009
Released on J-STAGE: November 03, 2011
JOURNAL FREE ACCESSThis paper presents a new modeling and control techniques for the steel sheet processing line based on a Hybrid Petri Nets (HPN) and Mixed Logical Dynamical Systems (MLDS) expressions. In the steel sheet processing line, the dynamics of sheets is represented by the differential (or difference) equation. On the other hand, some logical constraints such as change and/or inspection of the steel sheets are described by the discrete event-driven model. The proposed modeling method can harmoniously integrate these two different aspects, and can provide a suitable formulation for the optimization of the steel sheet processing line. Based on this hybrid system modeling, the closed-loop control can be realized by receding horizon scheme. Numerical examples demonstrate the usefulness of the proposed method.View full abstractDownload PDF (802K) -
Yuki YOSHIHARA, Yoshinari MAKINO, Nozomi TOMITA, Masafumi YANO2009 Volume 45 Issue 11 Pages 570-579
Published: 2009
Released on J-STAGE: November 03, 2011
JOURNAL FREE ACCESSHumans achieve reaching movements even when environment changes unexpectedly, by coordinating redundant joints of an arm in real-time. To realize such adaptability, we have proposed a new reaching model, in which instantaneous joint motion is determined based on local optimality of joint mobility. To study efficiency of our model, here, we compare our model to the minimum torque change model (Uno et al.,1989). Simulated results indicate that resulted trajectories by the two models are qualitatively almost similar, suggesting that a strategy to use a mobile joint dominantly is effective in generating a globally optimal motion.View full abstractDownload PDF (958K) -
Shun NISHIZAWA, Yu OHASHI, Keiji TATSUMI, Tetsuzo TANINO2009 Volume 45 Issue 11 Pages 580-586
Published: 2009
Released on J-STAGE: November 03, 2011
JOURNAL FREE ACCESSIn these days, a lot of productive activities are divided into small processes and coordinated under several collaborations. Therefore, we need to deal with large and complex scheduling problems. In this paper, we introduce a model in which several RCPSPs form a whole scheduling problem represented by a directed graph based on a precedence relation. Moreover, resource prices are getting higher recently and we must utilize resources more and more efficiently. In order to consider more efficient utilization of resources, we allow resources to be transferred from some RCPSPs to other RCPSPs under appropriate constraints in our model.
This paper addresses this problem of multiple projects with two types of objectives, minimizing the project completion time and minimizing the sum of weighted earliness-tardiness costs of each RCPSPs. For these objectives, we propose some methods to construct a schedule which has more efficient resource allocation. Our proposed methods have three phases. In the first phase we compute each RCPSPs' completion time, in the second phase we construct whole project schedule optimized for project completion time or total weighted earliness-tardiness costs and in the third phase we determine resource transfer schemes. We find out bottle neck RCPSPs of the schedule for each objective and improve their completion time by transferring resoureces to them from other RCPSPs in the third phase. We iterate these three phases for a certain number of cycles and obtain a better schedule. We also confirm efficiency of our proposed methods through computational experiments.View full abstractDownload PDF (431K) -
Tatsuo MOTOYOSHI, Hiroshi KAWAKAMI, Takayuki SHIOSE, Osamu KATAI2009 Volume 45 Issue 11 Pages 587-596
Published: 2009
Released on J-STAGE: November 03, 2011
JOURNAL FREE ACCESSThis paper proposes a logical framework that visualizes worker's system awareness. Generally, this awareness is difficult to compile in a educational book, and is seldom communicated in verbally. To show broad utility of system awareness, we need to show workers' conceptual structure of system operating.
We firstly carry out a work analysis of graph drawing task using a spreadsheet software. Based on it, we propose a logical representation of worker's system awareness using Formal Concept Analysis. This representation shows the workers' implicit rules which become the selection criteria of system functions. Furthermore, concept lattices which are obtained by this analysis, show the level of worker's applied skills of knowledge about system function. Finally, we discuss hierarchy of system awareness, and propose a way to find a new system usage from worker's experiences.View full abstractDownload PDF (1366K) -
Takeshi SHIBUYA, Shingo SHIMADA, Tomoki HAMAGAMI2009 Volume 45 Issue 11 Pages 597-604
Published: 2009
Released on J-STAGE: November 03, 2011
JOURNAL FREE ACCESSThis paper proposes two kinds of complex-valued Profit Sharing algorithm for learning in environments involving perceptual aliasings. These algorithms are implementations in complex-valued reinforcement learning framework. The CVRL framework aims to solve perceptual aliasing problem by using context. The CVRL is characterized by complex-valued action values and a internal reference value. The internal reference value is used to distinguish confused states and revised at each step. As another implementation in the framework, Q-learning has been proposed. Several experimental results have demonstrated the effectiveness of the Q-learning. However, Q-learning suffers from a disadvantage in parameter settings because of many parameters. Therefore, this paper focuses on Profit Sharing which has less parameter than Q-learning and proposes complex-valued Profit Sharing. Furthermore, this paper also describes a technique named multiple phase changes. This technique enables the agent to revise the internal reference value adaptively and realizes more sufficient learning in environments which have confused states at various intervals. Simulation results support that proposed method works well in such environments.View full abstractDownload PDF (869K) -
Takashi EDA, Nobutada FUJII, Toshiya KAIHARA2009 Volume 45 Issue 11 Pages 605-611
Published: 2009
Released on J-STAGE: November 03, 2011
JOURNAL FREE ACCESSIn product market with network externalities, outperformed products do not always prevail. Therefore, the product market is often modeled and examined by simulations to clarify those phenomena. In previous researches, multiagent system simulations in complex networks are often used and the feasibilities are confirmed. In this paper, it is proposed that threshold models are introduced into the multiagent system simulations in complex networks to consider consumers' heterogeneity. Computer simulations are conducted to verify the relationship between consumer network structure and heterogeneity affect product diffusion. In the results, it is revealed that consumers tend to purchase a product followed by the network externality effect although the consumers have little preference for buying a product, and consumers with high betweenness play an important role to product diffusion.View full abstractDownload PDF (888K) -
Ichiro KIMURA, Yasuaki KUROE, Hiromichi KOTERA, Tomoya MURATA2009 Volume 45 Issue 11 Pages 612-619
Published: 2009
Released on J-STAGE: November 03, 2011
JOURNAL FREE ACCESSThis paper develops models for “visual feature extraction” in biological systems by using “spiking neural network (SNN)”. The SNN is promising for developing the models because the information is encoded and processed by spike trains similar to biological neural networks. Two architectures of SNN are proposed for modeling the directionally selective and the motion parallax cell in neuro-sensory systems and they are trained so as to possess actual biological responses of each cell. To validate the developed models, their representation ability is investigated and their visual feature extraction mechanisms are discussed from the neurophysiological viewpoint. It is expected that this study can be the first step to developing a sensor system similar to the biological systems and also a complementary approach to investigating the function of the brain.View full abstractDownload PDF (753K) -
Noriaki KOUDA, Yuji KAWAI, Nobuyuki MATSUI2009 Volume 45 Issue 11 Pages 620-627
Published: 2009
Released on J-STAGE: November 03, 2011
JOURNAL FREE ACCESSRestricted Coulomb Energy(RCE) network is one of the competitive learning networks that are able to classify input data, together with the Self-Organizing Map (SOM) and the Learning Vector Quantization(LVQ). In the RCE network, there is no need for setting the number of required neurons before learning because the RCE network makes new neurons automatically to classify input data into correct categories. In this report, we propose a new RCE model and its network with RBF output function in order to reduce the number of neurons created in the network, and evaluate the classification efficiencies by solving the iris problem and the shuttle problem. Moreover, we investigate the abilities of the RCE model concerning the incremental learning unsolvable in LVQ. We find that our proposed model can perform in classification accuracies with smaller number neurons to compete with the conventional ones.View full abstractDownload PDF (1284K)
- |<
- <
- 1
- >
- >|