-
Shohei TANIGUCHI, Ryuta OZAWA, Ryota ISHIBASHI, Mitsunori UEMURA, Kats ...
2004 Volume 40 Issue 8 Pages
783-790
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
We develop a system to estimate joint torque of multi-joint structures such as humans and robots. The system consists of a parallel wire driven mechanisms which drives an end point of a multi-joint structure. The joint torque of the structures is estimated by measuring the end point force. We use only geometric relation between the end point and the joint for the estimation. In this paper, we describe a method to estimate joint torque, and verified precision of the system analytically. Lastly, we show some experimental results of estimation of human joint torque.
View full abstract
-
Takahiro FUJYU, Jun SUGAWARA, Hiroyuki TAKUNO, Haruki OKANO
2004 Volume 40 Issue 8 Pages
791-795
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
In portable falling weight deflectometer (FWD-Light) tests, the impact load and the deflection generated by means of a weight dropping on a loading plate are measured by the strain gauge type load cell and accelerometer, and are applied to nondestructive evaluation of the ground strength. In this paper, the techniques used in the load and the deflection measurements in the FWD-Light system are presented.
View full abstract
-
Experimental Evaluation of Reference Governors
Shinsuke OH-HARA, Kenji HIRATA
2004 Volume 40 Issue 8 Pages
796-805
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper considers an on-line reference management technique for a linear system with state and control constraints. The approach is based on the parameterized set of reference signals that avoid constraints violation. In result, the freedom of reference management is increased and the reference governor has good response characteristics. The proposed method is evaluated by simulation and experiment in the tracking control of the position servomechanism.
View full abstract
-
Takeshi HATANAKA, Kiyotsugu TAKABA
2004 Volume 40 Issue 8 Pages
806-814
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper presents two off-line design methods of a reference governor for constrained systems subject to a class of bounded disturbances. Both of these methods assure the infinite-time constraint fulfillment by settling the state variable into the maximal constrained positively invariant (CPI) set at the final instance of reference management. In order to improve the transient response, we introduce two performance indices based on
l1 and
l2 norms. Then, in both methods, the design of the reference governor is formulated as the minimization of worst-case performance under the constraint that the state should be settled into the maximal CPI set for any bounded disturbance in the prescribed class. It is shown that the reference governor designs based on the
l1 and
l2 norms are reduced respectively to a linear programming and a convex programming, which can be efficiently solved by existing numerical algorithms. The effectiveness of the present methods is demonstrated by a numerical example.
View full abstract
-
Mohammad Abdul MANNAN, Toshiaki MURATA, Junji TAMURA, Takeshi TSUCHIYA
2004 Volume 40 Issue 8 Pages
815-824
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper presents a minimal order rotor current and rotor flux observer for an indirect field oriented induction motor drive with consideration of core loss due to eddy current from the viewpoint of nonlinear observer using bilinear model. The state equations for an induction motor were derived which behaved as a bilinear system in terms of the product of input and state variables. The design of the proposed observer is based on Lyapunov's stability method whose estimation error converges to zero exponentially irrespective of the inputs. We have also proposed a control system by using multi-input and multi-output optimal regulator theory. The estimated rotor current and rotor flux are fed back to the multi-input and multi-output optimal regulator. Simulation results are presented to show the validity of the proposed controller as well as rotor current and rotor flux estimation of induction motor drive.
View full abstract
-
Hiroshi MARUTA, Manabu KANO, Hidekazu KUGEMOTO, Keiko SHIMIZU
2004 Volume 40 Issue 8 Pages
825-833
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Control performance monitoring is an important technology to keep highly efficient operation of production plants. Bad control performance is caused mainly by inadequate controller tuning or equipment malfunction. Valve Stiction is the most common problem in pneumatic control valves, which are widely used in the process industry. Stiction causes persistent fluctuation of process variables. Therefore, developing a method to detect stiction and distinguish it from inadequate controller tuning is crucial to help operators take an appropriate action for improving control performance. In the present work, valve stiction is modeled by taking into account its physical mechanism, and then new stiction detection algorithms are proposed. The usefulness of the proposed detection methods are demonstrated by comparing them with conventional methods. It is shown that the proposed methods can successfully detect valve stiction, distinguish it from bad tuning or disturbances, and quantify the degree of stiction, by using simulation data sets and real operation data sets obtained from several chemical processes.
View full abstract
-
Kazuto SHIRANE, Jun-ichi IMURA
2004 Volume 40 Issue 8 Pages
834-841
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper discusses from the algebraic points of view the relationship among the four properties of discretetime piecewise linear systems, that is, observability, determinability, mode estimatability, and mode-sequence estimatability. First, necessary and sufficient conditions for the system to be determinable, and to be modesequence estimatable are derived. Next, the observability property is characterized in terms of those conditions. Finally, the relationship between the observability property of the original system and the well-posedness property of the time-reversed system is discussed.
View full abstract
-
Approximation of the Reachable Sets for Bounded Inputs
Hidenori SHINGIN, Yoshito OHTA
2004 Volume 40 Issue 8 Pages
842-848
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Ellipsoidal invariant sets of linear discrete-time systems are treated to approximate the reachable sets for bounded inputs. All the ellipsoids are parameterized by a bilinear matrix inequality. The criteria for the ellipsoids are defined to represent the degrees of approximations. The optimal ellipsoids can be obtained by the convex optimizations with one scalar variable. The efficiency of the proposed method is demonstrated in a numerical example in comparison with a method of previous research.
View full abstract
-
Yuichi KOBAYASHI, Hideo YUASA, Shigeyuki HOSOE
2004 Volume 40 Issue 8 Pages
849-858
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Adaptive resolution of function approximator is known to be important when we apply reinforcement learning to unknown problems. We propose to apply successive division and integration scheme of function approximation to Temporal Difference learning based on local curvature. TD learning in continuous state space is based on non-constant value function approximation, which requires the simplicity of function approximator representation. We define bases and local complexity of function approximator in the similar way to the autonomous decentralized function approximation, but they are much simpler. The simplicity of approximator element bring us much less computation and easier analysis. The proposed function approximator is proved to be effective through function approximation problem and a reinforcement learning common problem, pendulum swing-up task and acrobot stabilizing task.
View full abstract
-
Hideyuki TSUKAGOSHI, Masashi SASAKI, Ato KITAGAWA, Takahiro TANAKA
2004 Volume 40 Issue 8 Pages
859-866
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper describes the numerical analysis and design for a higher jumping rescue robot using a pneumatic cylinder. First, the basic equations for jumping are derived and the simulation is performed. Then, the relationship between the jumping height and the pressure-receiving area of the cylinder is considered when the volume or the stroke of the cylinder is kept constant. This allows calculation of the optimal cross sectional area. In addition, the jumping height is also affected by the weight ratio between the rod and the cylinder tube. Based on these results, a robot equipped with a cylinder of the appropriate dimensions controlled by a well-selected valve is reengineered and demonstrated. Experimental results show that the improved robot can jump considerably higher than the former design with the same energy efficiency, as shown in the following video. http://www.cm.ctrl.titech.ac.jp/study/jump/home.html.
View full abstract
-
Izumi MASUBUCHI
2004 Volume 40 Issue 8 Pages
867-869
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
This paper proposes a generalized KYP-type matrix inequality that provides a necessary and sufficient condition of dissipativity of descriptor systems. While known bounded or positive-real inequalities are only valid for descriptor systems with particular realization of them, the proposed inequality condition is independent of the realization of descriptor systems. Thus the new inequality is used for analysis of descriptor systems with freely selected realization. Further, the proposed inequality is applicable not only to bounded and positive-realness but to dissipativity with respect to supply rates given by an arbitrary quadratic form of the input and the output.
View full abstract
-
Qiang WEI, Tetsuo SAWARAGI, Yajie TIAN
2004 Volume 40 Issue 8 Pages
870-872
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
Bin Packing (BP) problem is one kind of typical combinatorial optimization problems concerned about packing a certain amount of different items into the minimal number of
bins under specific constraints. In this paper, an organizational model called Garbage Can Model (GCM) is adopted to deal with a
relaxed online BP problems. In GCM, the organizational structure and fluidities of problems and resources provide many optimization and re-combination opportunities for arriving items in stream. Different from traditional online BP algorithms, which usually result in a less-optimal solution in pursuing an extremely online executing speed, GCM-based algorithm achieves finding a bounded optimal solution for a relaxed online BP in a
reactive fashion through seeking for a solution as optimal as possible as far as the time is allowed.
View full abstract
-
Takeshi MUTO, Yoshihiro MIYAKE
2004 Volume 40 Issue 8 Pages
873-875
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
We analyzed the Co-emergence process of cooperative walk between the handicapped person and the walking support robot "Walk-Mate". From the results, we suggested a relation between the mechanisms that characterize the Co-emergence process and those that improve handicapped gait motion.
View full abstract
-
Yohei KOBAYASHI, Yoichi NAGATA, Yoshihiro MIYAKE
2004 Volume 40 Issue 8 Pages
876-878
Published: August 31, 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS
The problem of time lag is critical in music ensemble. Several researches are reported to support ensemble under time lag by use special protocol or special music score. However these methods are not able to extend to more general ensemble. So in this research, we analyzed the affection of time lag on music ensemble. Under the condition of smaller time lag (less than 50ms), the ensemble was similar to the condition of no time lag. However, while time lag increased, one side of performer precedes the other side of performer, and finally the ensemble fails in the larger time lag (greater than 80ms). Therefore, the prediction of performance is necessary to perform the ensemble, and analysis of these mechanisms would be effective to construct the ensemble support system.
View full abstract
-
2004 Volume 40 Issue 8 Pages
e1
Published: 2004
Released on J-STAGE: March 27, 2009
JOURNAL
FREE ACCESS