This paper presents a development of the modeling of the human driving behavior based on the expression as Stochastic Switched ARX model (SS-ARX) focusing on the driver's collision avoidance behavior. First, the parameter estimation technique for the SS-ARX model is introduced based on the EM algorithm. Second, the parameter estimation technique is applied to the collected driving data, and find parameter set for each driving data. Finally, the performance of the SS-ARX model in the case of using as the recognizer is examined.

View full abstract

This paper concerns system identification of switched systemsin which each sub-system is expressed by a piecewise affine system.A considered system can be expressed by mixed logical dynamical system for which some design methods have been already known.In most cases, a number of sub-systems, that is a switched point is given ad hoc.However, it is not easy for multi-variable sytems.Therefore, in this paper, we propose an identification technique which takes into account a number of sub-systems and their paramteressimultaneously.The proposed method is applied to examples to show its effectiveness.

View full abstract

This study is about a novel small piezoelectric pump. The pump utilizes a thin vibration plate on which bimorph type piezoelectric actuators generate progressive wave. It doesn't use check valves and can send fluid to both forward and backward directions. In this report, principle of the proposed pump is described at first, and then examination by experiment is discussed. Motion of the vibration plates in the air for sinusoidal drive with phase difference among the bimorph cells is investigated. Characteristics of the pump are examined by experiments. As a result, effectiveness of principle of the pump with progressive wave is shown. Ability and problems of the experimental pump in the present condition are clarified.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

In blind source separation the definition of sources has an inherent indeterminacy; any linear transform of a source signal can also be considered another form of source signal. Due to this indeterminacy, the determination of the separator has a certain freedom. This paper proposes an idea for eliminating the indeterminacy, i.e., a principle for uniquely normalizing the separator. The output of the separator based on the principle undergoes the least distortion relative to the sources signals observed at the sensors. The obtained normalization has some good properties: easy implementation, no fluctuation, and no signal distortion.

View full abstract

[in Japanese]

View full abstract

In the case of the nonminimum-phase controlled object such as inverse response system,the feedback control could not improve the nonminimum-phase characteristics. Predictive controls with feedforward path have been designed by optimal or H infinity control. In this paper to compensate nonminimum-phase characteristics the timing and the amount of feedforward are derived with regards to minimum-phase state, where all-pass function corresponds nonminimum-phase. Numerical examples show that the feedforward control is efficient to improve the responses.

View full abstract

[in Japanese]

View full abstract

This paper presetns a design method for decoupled internal model control systems.

View full abstract

In industries controllers are conventionally and most commonly designed as a PID controller using a classical control method such as Evans' root locus method, etc. However, it is not easy to determine the three design parameters, i.e., proportional, integral and derivative gains, appropriately. Although design methods of a PID controller such as the ultimate sensitivity method by Ziegler and Nichols and its improved version by Kuwata have been proposed for some types of plant, engineers are often forced to do a tedious job of design iteration by trial-and-error.
In this paper, the author focuses on an I-PD type controller and proposes methods to determine its feedback gains based on the type-I optimal servomechanism. In an I-PD controller, integral action is applied to the control error and proportional and derivative actions to the output. Since the architecture of the control system is similar to the type-I optimal servomechanism, it is possible to design an I-PD controller from the gains of the optimal servo controller. The design procedure is summarized as follows. First the transfer function of a given single-input-single-output system is transformed into a state-space representation by choosing a state vector appropriately. Next a type-I optimal servo controller is designed for the system. Feedback gains of an I-PD (Integral-preceding Proportional Derivative) controller are determined from those of the servo controller. Furthermore, feedback gains of the I-PD controller provides those of a PID controller. Two design methods are presented according to the choice of the state vector. A controller determined from the feedback gains by the optimal servo controller is not necessarily an I-PD controller; in fact, it accompanies a filter unless the numerator of the transfer function is constant, and the derivative terms whose differential order is higher than one appears, when the plant order is higher than two. However, the resulting controller can be reduced to a typical I-PD controller by order reduction and gain adjustment. Numerical examples illustrate the effectiveness of the design methods and provide comparison with the results by the improved ultimate sensitivity method.

View full abstract

A set of stabilizing PID gains is examined by Bounding Crossing Theorem. First, the previous studies on the boundary of the stabilizing set are summarized where the conditions are represented by the frequency response of the plant. Next, a new method of determining the change of the number of unstable roots across the boundary of the stability set is given, by which the number of unstable poles can be easily determined for PI and PD planes.

View full abstract

In this talk, we provide a new method for tuning of parameters of a PID controller by using the least squares method applied to one-shot closed loop experimental data without system identifications.The method we give here has a practical advantage in the sense that what we require is only one-shot experimental data obtained from the closed loop and what we have to do is to perform the well known least squares method.Together with the proposal of a new method and algorithms, we dicuss the issues on the optimality of the obtained parameters and we give an illustrative example in order to show the validity of our result.

View full abstract

The final objective of this study is to realize an assist system for a human, especially a beginner, riding on an unicycle. At the fisrt step, this paper presents the modeling and stabilization control of the unicycle. The effectiveness of the proposed control system is evaluated by simulations.

View full abstract

In the spline method, one of the representative methods for PTP control of a robot arm, the designer should arbitrarily decide intermediate points of the path of the arm end and its velocity. This paper proposes a method which can decide the appropriate path uniquely by giving an area where the arm can move and by resorting it to a convex quadratic programming. This method designs a virtual target path in off-line with the use of singular vector of linearized dynamics of the robot arm.

View full abstract

[in Japanese]

View full abstract

We consider a problem of designing optimal smoothing spline curves by employing an approach based on linear control systems. In particular, we establish a design method of multi-level smoothing splines, a smoothing counterpart of Hermite interpolations, assuming that not only function values but also its derivatives are available as the data. The problem is formulated as an optimal control problem, and a necessary and sufficient condition for the existence of unique optimal solution is derived. The computational algorithm is simple as well as feasible numerically, and its effectiveness is verified by numerical examples.

View full abstract

[in Japanese]

View full abstract

Taking notice the independent parameters of Vehicle Speed (v) and Time-Headway (T) on a road traffic, measurement of both parameters was carried out, and discussed the variations of the measured parameters with respect to time (t-T,v) and T-v correlation. The flow (q) and density (k) were newly defined by T and v, and the traditional q-k, k-v, and q-v correlation are discussed. In concluding, we noticed that the k-v relationship might have some explicit relation, but the other ones had not enough information more than those in the t-T,v and T-v relationships.

View full abstract

[in Japanese]

View full abstract

We introduce a research of the "travel time prediction method" for the "traffic control system" of the expressway.In this paper, we introduce a research in the case of using the data that is obtained by infrastructure that is installed on the road, such as the vehicle detector, ETC system and so on.

View full abstract

[in Japanese]

View full abstract

In this research, a lane-changing assist system including surrounding vehicle recognition by vehicle-to-vehicle communication and an algorithm calculating appropriate lane-changing maneuver is presented. Comparing driving behavior of human drivers and the system, it is confirmed that the proposed system achieves similar or even superior performance to the human drivers.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

This study aims to synthesize a new driver assistance system by utilizing the vehicle yaw moment control which uses independent driving motors of electric vehicle. This paper presents the vehicle yaw moment control system which switches the mode of control according to human's driving task. The driving task recognition method is conducted by the analysis of the time history of steering wheel angle. The effectiveness of the proposed system was verified by computer simulation and experiments using micro-scale electric vehicle.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

This paper aims at developing a practical automatic parking technology for 4-wheeled cars, which is able to park cars subject to parking space constraint and steering angle saturation. The feedback law is designed based on the switching control method proposed by the authors for chained systems. The focus of this paper is to prove a condition on the initial state under which the stability is guaranteed even when the steering angle is saturated. After that, parking algorithms are proposed based on this condition. The proposed method is validated by numerous parking experiments.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

The purpose of this paper is to design the force/position hybrid ANN controller for tendon-driven mechanisms and to optimize the joint stiffness by GA. For compensation the nonlinearity of tendons and unknown parameters of robot dynamics, we proposed adaptive neural networks. Then, using Genetic Algorithms for optimization of joint stiffness to reduce the force error. We show the efficiency of proposed methods by simulation.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

In the reserch area of HDD control systems, it is believed that their heads are light and flexible structures. Thus, they are modelled it as two-mass systems. In this case, we showed a design method of SAC systems such that the we showed a design method of SAC systems such that the steady-state error of the proposed system is smaller than the ordinary system.

View full abstract

[in Japanese]

View full abstract

This paper proposes an architecture of vision-based robot systems which avoide obstacles and follow moving objects. Neural Network is used to predict position of moving object and to avoid the obstacle. We discuss effectiveness of the architecture from experiments using a robot arm.

View full abstract

The sensibility and the music knowledge are indispensable to decide an accurate chord when we create the chord progression with music. Recently, some chord progression can be created if there are two kinds of sources of a main melody and the accompaniment. However, it is difficult to decide an ideal chord progression only from a main melody because the combination of the chord exists a lot. Then, we propose a chord progression algorithm using Fibonacci sequence in this research.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract