The actual sound level fluctuates within a finite range and the observed data sometimes display level saturation owing to the existence of a definite dynamic range in our actual life. In this study, an evaluation method for sound environment is proposed from static and dynamic viewpoints in a form appropriate for the finite level range of the measured data. More specifically, in order to evaluate the sound environment around a main line, a static method to predict probability distributions of sound level at evaluation points based on the observation at a reference point is first derived. Next, a dynamic method to estimate the fluctuation wave form of a specific signal under existence of external noise. After introducing a Jacobi polynomial defined within a finite fluctuation range of sound level, by applying the well-known least mean squared method for the moment statistics with several orders, practical prediction and estimation methods are theoretically proposed. The effectiveness of the proposed theory is confirmed partially by applying it to actual traffic noise data at main line.

View full abstract

In this paper, we use the Baysian approach to evaluate the uncertainty of a measurement based on quantization data. We assume that the quantity before quantization obeys a normal distribution having the average μ and standard deviation σ. Based on a posterior probability estimated by n-times repetitive measurement, we are able to determine the probability density p (μ, σ) using the Baysian method. An uncertainty of measurement corresponds to a probability function concerning parameter μ. The probability function has been used to determine the estimation value after quantization and its uncertainty.
This paper describes the computational results for events such that the number of repeated measurement is smaller than fifteen and no more than three quantized values occur in each event. When all of the measurement data take the same value, the results show the conventional Type B evaluation described above results in underestimation of the uncertainty if the number of data is equal to four or less.
Furthermore, we show that estimations performed using this approach are significantly different from those performed by Type A evaluation when the event has indicating values for two ranged values.

View full abstract

In conventional inspection methods using ultrasonic sensors, the detection was performed by viewing the sensor's output with bare eyes, hence these methods had a low reliability. Also, the conventional methods often failed to detect multiple numbers of reflecting planes (such as cracks) which reside close to each other in the direction of depth. This is because the reflected waves from these planes are often overlapped, adding to that the reflected signal has a spindle oscillatory shape.
The present paper contemplates a method to simultaneously detect multiple numbers of reflecting planes using an ultrasonic sensor. Concretely, the sensor output is modeled based on the signal propagation model and a pattern matching technique is used to measure the propagation times of the reflected waves. Furthermore, an optimization method is developed to reduce the computation time significantly. Finally, experiments are conducted to verify the effectiveness of the proposed method.

View full abstract

In this paper, a dual problem of the motor control model based on the signal-dependent noise hypothesis is addressed. First, the motor control model including signal-dependent noise is reduced to an optimal control problem whose cost function contains a time-varying weight. Then an estimation problem under the time-varying output noise model is introduced as the dual problem of the motor control model mentioned above, and it is proven that this time-varying noise can be interpreted as a signal-dependent output noise. Finally it is shown by experimental results that the model of the estimation based on the signal-dependent noise does not qualitatively contradict to the estimation performed by human beings on some situations.

View full abstract

This paper presents a path following feedback control law for a wheeled locomotor referred to as a trident steering walker. This wheeled locomotor makes undulatory locomotion in which it transforms the rotations of its three joints into its movement by using four steering mechanisms. In this control law, we especially replace some mechanical elements of the wheeled locomotor with virtual mechanical elements and we also attach additional virtual mechanical elements to the wheeled locomotor. We prove that, by defining the virtual mechanical elements, it is possible to convert the kinematical equation of the wheeled locomotor into five-chain, single-generator chained form in a coordinate system where a path is a curved line axis and a straight line perpendicular to the tangent of the path is another axis. This conversion is more general because it includes a conversion performed in the Cartesian coordinate system where its two axes are straight and they are perpendicular to each other. Based on chained form, we derive a feedback control law which makes the wheeled locomotor follow any path as long as its curvature is two times differentiable. We have performed a path following simulation where the equilateral triangle base of the wheeled locomotor follows a Bezier path and its orientation relative to the tangent of the Bezier path converges into a desired angle. The validity of the conversion of the kinematical equation into chained form and the feedback control law is supported by computer simulations.

View full abstract

In this paper, a method for approximating the stabilizing solution of the Hamilton-Jacobi equation for integrable systems is proposed using symplectic geometry and a Hamiltonian perturbation technique. Using the fact that the Hamiltonian lifted system of an integrable system is also integrable, the Hamiltonian system (canonical equation) that is derived from the theory of 1-st order partial differential equations is considered as an integrable Hamiltonian system with a perturbation caused by control. Assuming that the approximating Riccati equation from the Hamilton-Jacobi equation at the origin has a stabilizing solution, we construct approximating behaviors of the Hamiltonian flows on a stable Lagrangian submanifold, from which an approximation to the stabilizing solution is obtained.

View full abstract

In control systems connected by networks with limited data rates, the question of interest is, what is the minimum data rate for given control objectives to be achievable? In this paper, we address the problem of guaranteeing the desired control performance for a linear time-invariant system where its sensor and actuator are connected by a finite data rate channel. We employ a dwell-time switching scheme, which is one of the techniques for preventing the problem of the chattering, and use a logarithmic quantizer as a quantizer. We design the controller that guarantees the control performance based on the quadratic cost function. Moreover, the relation between the data rate and the control performance is analyzed for the case with scalar systems.

View full abstract

In operating rooms of large hospitals in the developed countries, the problem of the shortage of scrub nurses in operation rooms has become serious. To compensate for this shortage, we have proposed Scrub Nurse Robot (SNR) system to assist a surgeon in exchanging surgical instruments during endoscopic and laparoscopic surgery. On design of the First SNR, motion capture system in an operating room to collect and analyze data of motions of a surgeon and a scrub nurse is constructed. Specifications of the First SNR are decided by consideration of the data of motion analysis. We evaluate exchanging motions of the First SNR using motion capture software and recognize that R-1 is faster than a human scrub nurse in helping a surgeon exchange an electric knife and a blood aspirator, which are used most frequently during operations, although R-2 is not so faster in the case of the other instruments. And we also recognize that the First SNR's minimum working space is 1.68 [m²]. Then, we develop 2nd SNR (ASULA) which is made its minimum working space small to 0.31 [m²] to be used in smalloperating rooms. ASULA is designed as one robot changing from 2 of the First SNR. ASULA recovers all of works that the First SNRs do, and adapts to the other functions, e.g. instruments cleaning system which is one of the most important task of scrub nurses. As a user interface of ASULA, voice recognition and reproducing system is installed. We construct a voice answering database. It is a data aggregate of words which are used in videotaped conversation between a surgeon and a scrub nurse during operations and which seems to be necessary. For the development the robot that exchanges surgical instruments with a surgeon, motion analysis to recognize a variety of data of a surgeon and a scrub nurse are important for unerring judgment for deciding the best specifications of the robot and the construction of the useful interfaces. Therefore, our development of SNR is one of the system integration pursuing a contribution for medical field by an assimilation of measurement field and mechanical design.

View full abstract

In the coexistence circumstance with humans and robots, sensory and emotional feeling of human should be considered when the robots interact to human.A typical unpleasant feeling at interaction is“pain”. It is difficult to separate the robots from human in the sense of time or place in coexisting environment unlike factory automation design. Thus a new concept of separation in the sense of safety is required.One approach is to design the controller based on the pain that is subjective of human. In this paper, artificial superficial pain model (ASPM) caused by impact is proposed. This ASPM model consists of mechanical pain model, skin model and gate control by artificial neural networks (ANNs). The proposed ASPM is applied to avoidance motion control of2-link robot and evaluated.

View full abstract

When an endpoint of manipulator is moving in touch with a rigid smooth surface, the manipulator should be controlled to achieve both a desired position trajectory and a desired contact force at the contact point. In this paper, we propose a new hybrid controller of repetitive learning control and model-based adaptive control for geometrically constrained robot manipulators. The convergence of joint angular errors and contact force error is proved under an appropriate initial conditions and the smoothness of the constraint surface. Furthermore, we show that the proposed hybrid controller can make a 2-link constrained manipulator track desired position and desired constraint force trajectoies accurately by using computer simulation.

View full abstract

Control systems composed of stable plant and integral control can be stabilized leaving no offset by small controller gain. This remains correct in the case where the feedback channel is randomly interrupted, provided probabilistic occurrence of the interruption is stationary and ergodic.

View full abstract

In the magnetic disk drives, the Initial Value Compensation (IVC) method at mode switching for head positioning servo is effective in improving the zero-input responses. This note presents a new IVC method which cancels the unfavorable poles of the closed loop by zeros, while minimizing a linear quadratic function of state variables.

View full abstract

A method to detect faults in plants by using an adaptive observer is proposed. The adaptive observer consists of a disturbance observer and a parameter adjusting law. The adjusting law is used to identify the uncertainties in the plant processes. A simulation result obtained to a water level control system is given to show the effectiveness of the propossed method.

View full abstract