-
Takeyoshi SUZUKI, Yayoi KANEKO, Oichi ATODA, Naoki MARUYAMA, Masaki TO ...
1999 Volume 35 Issue 2 Pages
167-175
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
A simple but reliable method is proposed for detecting existence of an object within a circular area with about 4m radius, which is optimal for logging trace of small wild animals. A beacon equipped with a magnetic dipole whose field is turned on and off according to its specific serial code is placed somewhere on the ground. If a magnetic sensor carried by an animal detects the field and find that the serial bit-pattern matches to one of predetermined codes, existence of the animal within the area around the corresponding beacon is known. It is shown theoretically that the radius of detectable area is very stable against changes of sensor height, sensor inchnation, and other electronic parameters since strength of dipole field is inversely proportional to cubed distance. A 14-bit code set including 622 individual codes are devised so as to decrease theoretical probability of spurious detection by introduction of Hamming distance and exclusion of bit-patterns with very little 1s. Chance of erroneous response to impulse and burst noise is also minimized. A trace-recording system is developed. In practical system the dipole coil is driven by AC in audio frequency band. A very low-power recording unit including a tiny sensing coil, an AGC amplifier, an analog comparator, a real-time clock, a static RAM, a micro CPU and lithium batteries is developed. The system is applied to research on a male badger. 65 beacons are deposited. During 40 days in late spring, 521 successive errorless records are acquired. Ecological analysis extracted five patterns of its nightly activities and revealed that it trips around its home range in every 3 to 4 days. Thus existence detection by coded magnetic field offers an effective instrument to animal ecologists.
View full abstract
-
Shogo TANAKA, Nobuyuki TAKEUCHI
1999 Volume 35 Issue 2 Pages
176-183
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Depending on the size and the weight of the load being transferred, various types of cranes are used in industries. Many attitude measurement and control systems for crane lifters have been developed. These systems are mainly aimed for suppressing the swings of the lifters when the trolleys come to a stop. To improve the efficiency and also not to damage the objects (loads), however, not only the measurement of attitude but also the exact recognition of the load position is important.
This paper proposes an automatic measurement system which combines fiber optical gyros (for attitude sensin) and range sensors and measures the position of cylindrical objects such as an iron-plate coil despite the swing of the lifter. The system makes use of the mechanism of the range sensor output to raise the measurement accuracy.
View full abstract
-
Koichi SAGAWA, Atsushi INA, Takayuki TAKAHASHI, Tadashi ISHIHARA, Hika ...
1999 Volume 35 Issue 2 Pages
184-190
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper describes an estimation method of several kinds of human moving behavior by using an acceleration signal and short term change of air pressure. The proposed method classifies the human behavior into normal walk, going up/down the stairs and going up/down in an elevator. The slight change of the air pressure during vertical movement is measured by a processing circuit constructed by using a Kalman filter which generates the signal corresponding to short term movement. The circuit also reduces noises caused by the change of weather and the internal noise. Periodic change of foot-to-head acceleration results from stepping action and the amplitude of the acceleration classifies human behaviors into a static state and a normal pace. The sign of the air pressure variation measured from the processing circuit determines the direction of vertical movement. A combination of these features gives a reliable estimation of human moving behavior. Level walk is estimated from the stepping action and the horizontal movement. Going up/down the stairs can be estimated from the stepping action and the vertical movement, while going up/down in the elevator comes from the static state and vertical movement. Several subjects are ordered to walk on the level ground, go up/down the stairs and go up/down in the elevator. The acceleration and the air pressure variation are measured and analyzed. Estimation results show that the proposed method is highly accurate to distinguish the human moving behavior at the rate of more than 95%.
View full abstract
-
Shogo FUKUSHIMA, Daisuke MORIKAWA, Hidekazu YOSHIKAWA
1999 Volume 35 Issue 2 Pages
191-199
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Accommodation, convergence and pupil constriction, which are called near triad, usually respond simultaneously at near reflex in natural view condition, but no research has been made on how pupil behaves following artificial near vision generated by binocular parallax.
Laboratory experiments were conducted on several subjects, to examine eye movement and papillary response under artificial binocular parallax, by using Eye-Sensing Head Mounted Display (ES-HMD). The ES-HMD can monitor the user's both eyes and calculate ocular information such as pupil size, eyeblink or eye movement, while presenting visual stimulus to his both eyes.
It was found that pupil constricts according to the increase of binocular parallax and vise versa, for the case of vergence normal subjects who have an ability of adaptation to binocular parallax, while conjugate eye movement instead of vergent eye movement appeared for the case of vergence anomaly subjects who would not succeed in binocular fusion. It was also found for a vergence normal subject that continuous periodic pupillary movement of constriction and dilation has tendency of hysteresis, which is caused by different muscle organ between constriction and dilation. Another results also suggest that the direction of smooth pursuit eye movement in a vergence anomaly under binocular parallax depends on the dominant eye.
View full abstract
-
Jinglu HU, Kousuke KUMAMARU, Katsuhiro INOUE, Kotaro HIRASAWA
1999 Volume 35 Issue 2 Pages
200-207
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper deals with the problems of robust fault detection using Kullback discrimination information (KDI) in presence of nonlinear undermodeling. The systems to be diagnosed are assumed to contain certain unknown nonlinear elements. The fault detection is performed by applying the KDI to a linear ARMAX model with model uncertainty, in which error due to nonlinear undermodeling is described using a group of fuzzy models with adjustable parameters. The estimate of modeling error is considered in the KDI analysis and thresholding decision for robustness realization. The effectiveness of the proposed robust fault detection scheme is examined through numerical simulations.
View full abstract
-
Kazunori YASUDA
1999 Volume 35 Issue 2 Pages
208-212
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper is concerned with the quadratic stability and the quadratic stabilizability via semi-state feedback of a linear descriptor system with time-varying and norm-bounded uncertainties which lie in every coefficient matrix of the system. First, it is shown that the uncertain descriptor system under consideration can be equivalently transformed to the system without uncertainties in the coefficient matrix for the derivative of the descriptor variables, and the conformability of the transformed descriptor system is introduced, which corresponds to the property for a time-invariant descriptor system to have a unique solution and no impulsive modes. Secondly, the quadratic stability is defined and then a necessary and sufficient condition for the quadratic stability is provided based on the generalized Lyapunov theory. Finally, a necessary and sufficient condition for the quadratic stabilizability and a stabilizing controller are given. æ
View full abstract
-
Shigeru YAMAMOTO, Masataka KIMURA, Ryozo NAGAMUNE
1999 Volume 35 Issue 2 Pages
213-220
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
In this paper, we propose a method of identification for a discrete-time state-space model set. The identification problem is to determine the parameter set which characterizes the model set such that the model set is consistent with given input/output data in the time-domain. In addition, it is required that there exists a stabilizing feedback controller for the model set. These consistency and existing conditions are derived in terms of linear matrix inequalities. The proposed algorithm iteratively solves these LMI's to obtain the model set and the controller. A remarkable result shows that there is no difference between the Frobenius norm and spectral norm which measures the bound of parametric uncertainty of the model set in the context of identification and control in this paper.
View full abstract
-
Shiro MASUDA, Akira YANOU, Akira INOUE, Yoichi HIRASHIMA
1999 Volume 35 Issue 2 Pages
221-229
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Generalized predictive control (GPC) can be derived in the state space approach as well as in the polynomial approach, and several researches which give relations between these approaches had been discussed. However, strict equivalence between GPC in the polynomial approach and one in the state space approach has not been shown so far. Therefore, in this paper, GPC in the state space approach with reduced order observer is proposed, and the proposed method is shown to be exactly equivalent to GPC in the conventional polynomial approach. From the result of the paper, the quantitative relation between these approaches can be clarified.
View full abstract
-
Shinsuke TAKAHASHI, Tohru WATANABE, Teruji SEKOZAWA
1999 Volume 35 Issue 2 Pages
230-237
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Environmental concern has recently led to severe regulations on exhaust emissions and fuel consumption in automotive engines. Direct fuel injection systems have attracted attention as new technology for meeting these regulations. This system needs exhaust gas recirculation in order to reduce NOx components on exhaust emissions.
This paper presents an exhaust gas recirculation model for control design and evaluation in direct injection systems. The proposed model is constructed based on physical laws such as the mass conversation law and the ideal gas equation. This enables the model to simulate the engine process accurately.
Comparison between simulation results and experimental results shows that the model has high accuracy. Moreover, a few examples in which the proposed model is applied to control design are shown.
View full abstract
-
Yoshihiro MATSUI, Tomoyoshi TAKEUCHI
1999 Volume 35 Issue 2 Pages
238-243
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Vibration suppression control of resonant mechanical systems is one of important problems on motion control. A control for a simple three inertia model of such resonant mechanical systems is given by the SICE as a benchmark problem. Three-inertia systems have two resonant modes. First, we design a PID speed controller in order to suppress the vibration of the first resonant mode of the three-inertia system. Next, we design a position control system using the PID speed controller. The system consists of two control loops. The inner speed control loop includes the PID controller and the outer position control loop includes a proportional controller. However the system is unable to suppress the vibration of the second resonant mode. Therefore we add a phase-lead compensator to the inner control loop of the system to suppress it. It is clarified that the proposed control system is useful for three-inertia systems by computer simulations.
View full abstract
-
Naoyoshi YUBAZAKI, Jianqiang YI, Kaoru HIROTA
1999 Volume 35 Issue 2 Pages
244-252
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
“SIRMs Dynamically Connected Fuzzy Inference Model” is proposed for plural input fuzzy control. For each input item, a SIRM (Single Input Rule Module) is constructed, and a dynamic importance degree is defined by a sum of a base value and a dynamic value. The base value is used to insure the role of the input item at steady state, while the dynamic value is designed to change with control situation so that the importance degree is tuned in real time. At each sampling time, the model output is obtained by summarizing the products of the dynamic importance degree and the fuzzy inference result of each module. The concrete design method of the proposed model is also given when constant value control plants are taken into consideration, and it is made clear that each dynamic value can be determined just based on the current local information of the corresponding input item. Furthermore, the proposed model is proved to be nonlinear PI control when it adopts the output error and the first-order change in the error as its input items to control first-order plants, and nonlinear PID control when it adopts the output error, the first-order change in the error, the second change in the error as its input items to control second-order plants. The proposed model is applied to typical first-order lag plants and second-order lag plants. The simulation results show that the reaching time can be reduced without any steady-state error, overshoot, or vibration compared with the conventional linear PI or PID controller.
View full abstract
-
Zibo KANG, Seizo FUJII, Chaojun ZHOU, Kazuya OGATA
1999 Volume 35 Issue 2 Pages
253-261
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
An Adaptive switching control scheme is proposed for a planar gantry crane. The length of the crane cable is subject to change in different operating conditions. Based on this control scheme, the control to the planar gantry crane can be switched among several controllers, each designed for a different fixed-length nominal model. Each “fixed-length controller” consists of double loops, one of which is a state feedback controller, and another one is a nonlinear controller aimed to compensate the unknown friction. An “unknown-input observer” is designed to estimate the state variables in order to avoid the influence of the unknown friction. The stability of each “fixed-length controller” is guaranteed by Lyapunov's direct method. Experimental results are presented to demonstrate the effectiveness of the proposed control scheme.
View full abstract
-
Toshiyuki KONDO, Akio ISHIGURO, Yoshiki UCHIKAWA
1999 Volume 35 Issue 2 Pages
262-270
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Conventional artificial intelligence (AI) system has been criticized for its brittleness under dynamically changing environments. Therefore, in recent years much attention has been focused on the reactive planning approach such as behavior-based AI, new AI, animat approach and so on. However, in behavior-based AI approaches, the arbitration among competence modules is still an open question. On the other hand, biological information processing systems have various interesting characteristics viewed from the engineering standpoint. Among them, the immune system plays an important role in maintaining its own system against hostile environments. Based on this consideration, we have been investigating a new decentralized consensus-making system for the behavior arbitration of autonomous mobile robots inspired from the idiotypic network hypothesis in immunology. In this paper, we propose a new reinforcement learning method using advantage of the proposed network architecture. To confirm the validity of our proposed method, we carried out some simulations.
View full abstract
-
Tadashi HORIUCHI, Akinori FUJINO, Osamu KATAI, Tetsuo SAWARAGI
1999 Volume 35 Issue 2 Pages
271-279
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Reinforcement learning is an essential class of machine learning for autonomous agents to acquire adaptive and reactive behaviors.
Q-Learning is a widely-used reinforcement learning method which deals with only discretevalued inputs (states) and outputs (actions). In this paper, we propose a new method of
Q-Learning where fuzzy inference is introduced to represent
Q-function (action value function) that evaluates state/action pairs, in order to to deal with continuous-valued inputs and outputs. Within this method, the steepest descent method is utilized to update
Q-function so that the parameters of fuzzy rules are tuned both in antecedent part and consequent part. Furthermore, we extend this method to the case of discrete actions by preparing a fuzzy inference system for each action in order to realize the speed up of learning. Finally we show the effectiveness of this method by comparing it with other methods through the applications to control problems such as the cart-pole balancing problem and the ship navigation problem.
View full abstract
-
Hiromitsu KATO, Hiromitsu KURISU, Teruji SEKOZAWA, Nihei TACHI
1999 Volume 35 Issue 2 Pages
280-287
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
We propose a hybrid method using interactive multiple criteria programming and minimum cost flow analysis to make a useful schedule for waterworks. In case of constructing a mathematical model, it is difficult to adjust the model parameters in the actual application, and to make the user satisfied with the solution under the predetermined parameters. In our research, the parameters are generated automatically based on the abstract plan, in which interactive multiple criteria programming is employed, and then the minimum cost flow is calculated under the generated parameters.
Flexible water supply scheduling is indispensable to assign water using the limited facilities efficiently in various situations. Traditionally, it is a common approach to formulate the minimum cost flow problem to make a schedule. However, good modeling is an essential premise to make a useful plan to distribute water, and elaborate work is necessary for good modeling. It is also difficult to adjust the solution, reflecting the user's aspiration. Multiple criteria programming(MCP) has been proposed to overcome these disadvantages of single-objective optimization. Nevertheless, it is not quite effective to apply MCP to large-scale problem like water supply scheduling because of complexity and formulation difficulty. Using our proposed method, the whole problem can be divided into two parts. In abstract planning, interactive multiple criteria programming is applied to extract the user's request for the intake flow amount by using aspiration levels. Then, cost coefficients of the multi-layered network model are generated automatically based on the abstract plan. Since the minimum cost flow is defined as the flow which can nearly reproduce the abstract plan and minimize the deviation of flow rate, resulting minimum cost flow can satisfy requirements of both reservoir level recovery and flow smoothing. As a result, the proposed method can make an applicable schedule quickly, reflecting the user's aspiration flexibly.
View full abstract
-
Koichi YAMADA
1999 Volume 35 Issue 2 Pages
288-293
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper addresses a reasoning in a hierarchical causal network that derives conditional possibility distributions of arbitrarily chosen nodes, when some other nodes are instantiated. The study is a step toward a more general reasoning that works on a directed acyclic graph in a possibilistic way. However, it is ahead of some studies done before which dealt with the inverse problem of causation using Possibility Theory, because the causal reasoning in this study includes the inverse problem.
First, the paper introduces Causation Event expressing an event that “a cause
actually causes an effect”, and Conditional Causal Possibility meaning conditional possibility that a causation event occurs when its cause is observed. Then, it discusses some characteristics of conditional causal possibilities and relations with conventional conditional possibilities. It also applies them to causation analysis which is a problem to derive conditional possibility distributions of arbitrarily chosen nodes in a hierarchical causal network. Finally, it shows a numerical example and discusses its results.
View full abstract
-
Masakatsu KAWATA, Naomitsu SHIMAZU, Kazuo INOUE
1999 Volume 35 Issue 2 Pages
294-296
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
We have already proposed a novel numerical approach to compute the approximate solution of the Hamilton-Jacobi equation in the nonlinear
H∞ control problems. In this paper, our proposal is applied to an inverted pendulum, and its validity is shown through experiments as compared with the linear
H∞ control.
View full abstract
-
Tadashi EGAMI
1999 Volume 35 Issue 2 Pages
297-299
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
A new design method of low-degree optimal preview repetitive control system is proposed. In this method, the degree of Riccati equation can be reduced to that of the controlled object and the initial response can be improved largely by the effort of the preview feedforward compensation.
View full abstract
-
Yukio MURANAKA, Eitaro AIYOSHI
1999 Volume 35 Issue 2 Pages
300-302
Published: February 28, 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper presents a new approach to solve line-balancing problems using the discrete and linked state transition neural networks whose feasibility for the constraints is kept by cooperative and simultaneous change of the neurons states. The simulation results for the simple line-balancing problems demonstrate more effective global convergence than Hopfield's neural networks
View full abstract
-
1999 Volume 35 Issue 2 Pages
e1
Published: 1999
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS