In recent years, it has been expected that biped robots walk on level ground with an effective use of their physical structure. The present paper proposes a biped robot which utilizes a torque for holding its torso and for swinging its arm as a means to the walking. Our numerical simulation shows that this robot can walk efficiently on level ground over a wide range of speed.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

Human support system is attracted by support persons. Many support systems are used position control and velocity control. These systems move a human mandatorily. This paper proposes method of human support system using motion coping system, The motion coping system saves human motion position and force. The proposed method supports the human based past individual motion. Effectiveness of the proposed method is verified by experimental results.

View full abstract

We have already proposed a handshake approaching motion model based on the analysis of handshake motion between humans. Furthermore, the effectiveness of the proposed model is demonstrated by sensory evaluation. However, in handshake motion between a human and a robot, an influence on humans depending on size and shape of robots, is not clarified. Therefore, we develop a handshake motion analysis system by the mixed reality technology using a robot model arranged on a virtual space and a handshake robot arranged on a real space. Using this system, humans can shake hands with many types of robots by superimposing a CG image on the actual robot.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

This study is concerned with the synthesis of periodically time-varying memory state-feedback controllers (PTVMSFCs) for discrete-time linear systems. In our preceding studies, we have already established basic strategies for LMI-based (robust) H-infinity PTVMSFCs synthesis, and in this paper we extend those results to the H2 performance criterion. In the H2 case, the main difficulty stems from the fact that we have to ensure the existence of a common solution for multiple LMI conditions that are related to Lyapunov inequality and inequalities for bounding traces of Lyapunov matrix. We overcome this difficulty by deriving an analytic representation of LMI solutions and derive a necessary and sufficient LMI condition for the H2-PTVMSFCs synthesis. Based on these results, we also consider robust H2-PTVMSFC synthesis for polytopic-type uncertain LTI systems.

View full abstract

In this paper, we propose an exterior-point approach for finding feasible solutions of the reduced-order output feedback H-infinity control problem and show its application to the mixed H-infinity/D-stability control problem. The advantage of our approaches is to be able to use a non-stabilizing controller as an initial point. Numerical examples show the efficiency of our approaches.

View full abstract

The dilated LMI characterizations enable us to use parameter-dependent Lyapunov functions for robust system analysis and synthesis problems. Based on the dilated LMI characterizations, a successive linearization of bilinear matrix inequalities (BMIs) are proposed. This paper demonstrates the effectiveness of this linearization technique by applying to AC motor control system synthesis.

View full abstract

A probabilistic cutting plane technique based on maximum volume ellipsoid center is presented for solving a robust optimization problem which is to minimize a linear objective function subject to a parameter dependent convex constraint. The algorithm employs random samples of the parameter and stops in a finite number of iterations giving a suboptimal solution and a suboptimal value as outputs. It is shown that the number of iterations is of polynomial order of the problem size.

View full abstract

Many robustness analysis problems admit a formulation as a robust semidefinite program (robust SDP) with linear fractional representation (LFR). Robust SDPs, which involve infinite number of inequality constraints, are reduced to a standard finite-dimensional SDP (LMI optimization problem) through a multiplier. Though the optimum of the latter in general merely provides an upper bound of that of the robust SDP, they often coincide with each other, or the upper-bound relaxation is {\em exact}, for problem instances. Verification of exactness has been receiving considerable amount of attention recently, which allows us to decide that an upper bound is accurate enough and stop updating multipliers to improve the upper bound with increasing computational load. Among several algorithms to certify exactness, a recursive method has been proposed which is applicable to upper-bound relaxations derived through general multipliers, not specific ones such as D-G scaling or SOS relaxations, where the LFR in robust SDPs are assumed to have repeated-scalar blocks. This paper generalizes those results to handle robust SDPs and upper-bound relaxations without the restriction on blocks in LFRs to be repeated-scalars.

View full abstract

An approach to a robust semidefinite programming problem is presented. This is an optimization problem whose constraint is a linear matrix inequality nonlinearly dependent on uncertain parameters. It covers various control problems having a nonlinear structure. In particular, a matrix-dilation approach to this problem is presented and its properties are discussed. Also, stability analysis of an aperiodic sampled-data control system is considered as an example of a control problem with a nonlinear structure. A new technique is introduced for exponential parameter dependence possessed by this problem.

View full abstract

This paper proposes analysis and synthesis methods of dynamic quantizer for linear feedback single input single output (SISO) systems with discrete-valued input in terms of invariant set analysis. First, we derive an optimal dynamic quantizer within the ellipsoidal invariant set analysis. In the case of SISO minimum phase plant, next, we present that the structure of our proposed dynamic quantizer is also optimal in the sense that the quantizer gives an optimal output approximation property.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

To obtain a near optimal control policy for real world chemical processes, in this paper, we focused our attention on a new meta-heuristic method termed differential evolution (DE). Compared with conventional approaches characterized by variational logic as well as by the inconveniences of simultaneous optimization methods for differential-algebraic equation, we have shown DE's ability to derive a near optimal solution adaptive to various requirements in practice. Since the algorithm is straightforward and flexible to manage various conditions, we can cope with high dimensionality and outstanding non-linearity peculiar to chemical process in real world applications. Taking reaction processes, we applied the proposed method under various meaningful conditions, and validated its adaptability in comparison with other methods.

View full abstract

This paper introduces the concept and the design procedure of nominal characteristic trajectory following (NCTF) controllers and the performances of control systems with the controllers. A basic NCTF controller comprises a nominal characteristic trajectory (NCT) defined as a reference motion trajectory on the phase plane and a PI compensator. The NCTF controller design does not require known-model parameters and the knowledge of control theory. Three kinds of NCTF controller were designed for conventional ball screw mechanisms having friction characteristics and their positioning, tracking control and contouring control performances were examined.

View full abstract

Since natural disasters are reported in various regions, an advanced ubiquitous monitoring system for disaster prevention is proposed. The system is developed to detect early disasters of constructions along expressways. Mass low cost sensors are required for monitoring such big constructions. Therefore, acceleration sensors made by MEMS technology are developed for monitoring the deformations of structures such as the landslide of slopes. Each sensor is connected with a node of the ubiquitous network. The nodes are connected each other by wireless correspondence. They organize a communication network, autonomously. A node is connected with the internet. When any change is found, warning signal is sent to the head office and furthermore to the administrators by portable phones. The accuracy of the sensors is investigated. The practical test using twelve nodes are examined for the slope of horizontal length of 100 m and height of 40 m along the expressway, for one year. Power is supplied by a battery (Unit II type) for each node during four months. It is verified that the system is significantly effective for the perspective of disasters.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

In this study, an automatic diagnostic system for a rotor shaft defect of DC motor is developed. Oscillatory data induced by rotation of motor shaft is assembled using oscillation sensor imitating human usage tool for oscillation sensing. Wavelet transformation and Self Organizing Map are employed to analyze the assembled acoustic sounds. To check the validity of the proposed method of diagnosis, experiments are made using a small DC motor and an actual high powered motor. As the results, acoustic sounds are classified according to its degrees of defect.

View full abstract

A small sized AGV is developed to check the applicability of circuit design method.To minimize the occupied area of the whoile circuit,objective functions for the design are compared.To derive the solution,optimzation program is developed based on SA method.The running test of the designed circuit is carried out to check the realization of the desired performance.

View full abstract

This paper proposes new type of steering control system to follow in the same wheel track for an articulated cargo train system to reduce the phisical distribution cost. Proposed system is supported all electrical traction trainn system whichi is incliusing high level automatic train controlsystem,high level train following system. detailed mechanism and systems are proposed and discussed.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

This paper address an routing problem for Automated Guided Vehicles used in the Flexible Manufacturing Systems. FMS which supposed in this study has one-way transportation path, and an AGV cannot get ahead of the other. We suppose that the transportation request arise from the production schedule. Considering path constraints and the production schedule, we propose a routing method which can make feasible and least-cost routes.

View full abstract

It is difficult to adjust the weight parameters of a multiple index on multiobjective product-mix problems. The solution is often far away from the schedule that is wanted. Therefore, to obtain a desirable schedule we study an approach that makes the schedule come to the intended target level of each index. To confirm the effectiveness of this approach, the linear goal programming model due to goal vector is verified by numerical experimentation with the number of products and the total profit as the index. As a result, we can make the solution come to the target levels of their index in a balanced manner.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

In order for humans and robots to engage in direct physical interaction several requirements have to be met. Among others, robots need to be able to adapt their behavior in order to facilitate the interaction with a human partner. This can be achieved using machine learning techniques. However, most machine learning scenarios to-date do not address the question of how learning can be achieved for tightly coupled, physical touch interactions between the learning agent and a human partner. This paper presents an example for such human in-the-loop learning scenarios and proposes a computationally efficient learning algorithm for this purpose. The efficiency of this method is evaluated in an experiment, where human care givers help an android robot to stand up.

View full abstract

[in Japanese]

View full abstract

Our goal is to clarify the role of embodiment in the early stage of the development through experiments with the musculoskeletal infant robot we have built up. This paper describes the way for the infant robot to acquire crawling locomotion utilizing central pattern generator and online optimization algorithm referred to as Powell's method. A coordinated motion of the musculoskeletal infant robot is successfully learned so that it can go forward after the optimization.

View full abstract

We propose a situation-dependent predictor which can change prediction depending on the object's shape and the environment. In order for a robot to obtain the concepts by itself, we propose an attention model with which a robot decides what to attend and when to serch a new object.

View full abstract

[in Japanese]

View full abstract

Real parallel computing with a quantum computer attracts much research interests due to its extreme high potential. Many quantum computation algorithms utilizing quantum mechanical behavior such as superposition of quantum states have been proposed. Recently, Farhi et al. have proposed an adiabatic quantum computation (AQC), which can be applied to various problems including non-deterministic polynomial time problems if one can know an appropriate Hamiltonian for a target problem. We have proposed a neuromorphic adiabatic quantum computation (NAQC) as the AQC with energy dissipation and an efficient method for designing a final Hamiltonian in consideration of the analogy with a neural network. In this paper, we show preliminary results for large scale quantum bit network with NAQC and we evaluate results for an autoassociative memory in the network.

View full abstract

[in Japanese]

View full abstract

In this paper we describe comparisons between bifurcation phenomena of a resonate-and-fire-type digital spiking neuron and an analog spiking neuron. We can compare bifurcation phenomena using return maps which describe the neurons' dynamics. We also propose an FPGA implementation method of the resonate-and-fire-type digital spiking neuron and confirm its operation experimentally.

View full abstract

In this paper, we introduce a resonate-and-fire type spiking neuron model and analyze its stability of synchronization phenomena to an input spike train. We also introduce a parameter update rule to memorize a period of input spike train and analyze its pull-in process. Then we introduce a pulse-coupled network of the neuron model and show that the network and a Kohonen's SOM have similar functions.

View full abstract

[in Japanese]

View full abstract

In this manuscript, we discuss the notion of "robustness" in the context of statistical learning. We focus on the paradigm of "learning from examples," and firstly review basic concepts of robust statistics within the framework of unsupervised learning, where we discuss that there is some similarity between robust statistics and robust control theory. We then review generalization in the context of supervised learning, and discuss that problems of generalization can be regarded as arguing some sort of problems of "robustness."

View full abstract

In this work, we focus on the problem of Channel State Information (CSI) feedback by Uplink (UL) Random Access (RA) in an Orthogonal Frequency Division Multiple Access (OFDMA) system. We introduce the concept of variable collision protection, where the probability that a feedback information experiences a collision depends on its importance. By partitioning the CSI into orthogonal layers of priority, and by allocating different numbers of RA slots to each layer, this scheme ensures that the feedback success probability is higher for the CSI with better quality, as it is more likely to be used by the scheduler. Analytical and simulation results show that our proposed scheme provides an excellent trade-off between system performance and the amount of feedback overhead.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract

When objects of embedded systems are enlarged to include human it is necessary to collaborate between natural scientists and social scientists in order to develope such embedded systems.

View full abstract