Control System
Control Theory Specialized for Periodic Motion
In our daily life, we often experience periodic motions
such as walking or bicycling. There exists a design
method for such periodic motions effectively by
making use of time delay of signals cleverly.
By completing such a technique as periodic control theory,
we are trying to realize various applications like
stabilization of passive walking robots and power assist
control of electic bicycles. We also study remote control
systems robust to transmission delay or data loss, by
using the structure obtained by the above theory.
Constrained Control
In mechanical control systems, we have to take various
types of constraints (limitation or saturation)
into consideration. For example, in order to move a motor
of a robot, we cannot apply inifinitely large voltage.
In conventional linear control theory, however, such
a dangerous controller may be realized. In recent years
these constrained systems have been extensively studied
and various good design methods have been proposed.
In our lab, we study practical control methods for
constrained systems, by considering not only the existence
of the constraints explicitly but also a kind of
optimality from an engineering viewpoint.
Adaptive and Learning Control
In adaptive and learning control systems, we improve
control performance while operating (in real time),
by means of various techniques of learning or adatation.
This is closely related to system identification.
Currently we are developing methods such as feedback error
learning, blind identification, and disturbance rejection.
We will further tackle the problem of closed-loop
identification and adaptive signal processing.
Communication System
Next-Generation Photonic Network Architecture
By using WDM (Wavelength Division Multiplexing) technology, Tera-bit data transmission
can be performed in photonic networks. Therefore、in order to create an ubiquitous society,
it is necessary to construct photonic networks by using WDM technology.
In this research, we study photonic network architectures in order to transmit data
effectively, and we aim to propose a new powerful architecture which is used across the globe.
Capability-Aware P2P system for Unibiquitous Computing Environments -Simulation, Development, and Implementation-
We propose, develop, and implement a new large-scale P2P system for ubiquitous computing environments
In the future, several types of nodes such as high-performance computers, sensor nodes, and mobile phones
will participate a P2P system.
In this research, we proposed a new capability-aware P2P system based on skip list and try to implement this system with HP iPAQ.
Dynanmic Frequency Allocation in IEEE802.22 WRAN
Recently, IEEE802.22 Wireless Regional Area Networks are promising as a new wireless broadband access network
for rural areas. In IEEE 802.22, software radio technology is utilized in order to utilize
limited frequencies.
In this research, we proposed a new dynamic frequency allocation method and evaluate its performance by simulation.
Now, we consider a ranking-based frequency allocation, and we will receive fund from mc Access Kinki from 2008 to 2011.
Network Business Management
In this research, we study network business strategy in order for carrier or
Internet service provider (ISP) to grasp good business opportunity of network services.
By considering pattern of user activity, network resource, network service cost,
and quality of service (QoS), we aim to provide optimal business strategy with operations
research or game theory.
Control & Communication
Remote Control Over Networks
Recent progress in network technology has made it
possible to construct control systems over network.
This is now one of the hot issues in control engineering.
It is known, however, that various network phenomena
(transmission delay, packet loss, torsion, error, etc) are
hard to treat in convensional linear control theory.
In our lab, we are trying to develop a new methodology
for remote control over network, mainly using an actual
remote control system constructed as a test bed.
Networked Control Systems
Some practical situations require stability to be never
violated, where networked control should be used carefully
in this regard. We are tackling this issue by
distributing controller functions and by localizing
some loops subject to delay or loss. This is done by
means of the LFT structure of stabilization or H infinity
suboptimal controllers.
We have already performed experiments with a proto-type
using RT-Linux.
Synchronization of Multiple Equipments over a Network
In this research, we develop methodology for cooperative works of multiple equipments over a network. To demonstrate the effect of our proposed switching method for network induced delay, we perform experiments with plants modeled on mechanical tools. Right and left movies show the experimental results in the case without/with proposed method (switching observer). Signals from the central motor are transmitted to the right motor. From these results, it can be confirmed that our observer compensates for the effect of random delay and all motors are synchronized.
Optical Grid Management and Optical Traffic Management with Control Theory
In this research, in order to manage both computing and network resources effectively
in optical grid networks, we propose a dynamic management method which is implemented
in a job manager. The proposed method adjusts the number of
tasks which are transmitted to a computing site with PID control
so that a resource manager can store a constant number of
tasks, resulting in an effective utilization of computing resources.
Moreover, lightpaths are established and released dynamically by
comparing output signal of PID controller with the maximum
number of transmitted tasks, resulting in an effective utilization
of network resources. We evaluate the performance of the
proposed method with Matlab.
Then, we try to management traffic in optical networks with PID control.
In the future, we extend the above methods so that Predictive Functional Control can be utilized.
Sensing & Signal Processing
Visual Feedback
自動車やホバークラフトなどのラジコンを用いた制御実験を行う場合，
ラジコン自体は数万円までで購入できますが，ネックになるのはセンサとくに天井カメラ等から
対象物の座標を検出するカメラシステムで，ミリ秒程度のデータレートを要求すると
非常に高価なります． 汎用のUSB カメラを用いて30フレーム/秒の処理システムを試作し
ボールアンドビーム装置のビジュアルフィードバック系を試作しました．これをさらに発展させ，
複数センサの協調による高速化・高精度化や複数のセンサ情報を統合する際のキャリブレーションや
フィルタリングの問題などを扱いたいと思います．
Moving Robot Control over Sensor Networks
In this research, we develop moving robot control system over sensor networks.
We use e-puck as a moving robot and mica mote as sensor nodes.
The robot and sensor nodes is extended by modifing Tiny OS (nesC program).
Currently, we look for researchers and students for this project.
ICA
In this research, we study system identification, fault detection, and disturbance rejection via independent component analysis.
Contents 