Description of Projects

Theory of Nonlinear Robust Control

Robust control is an important concept of advanced control which takes into account mismatches between mathematical models and true behavior of physical systems in analysis and design of control. The mismatches are considered as ``uncertainty'' in systems. The last two decades have seen the emergence of sophisticated mathematical techniques for dealing with robustness with respect to such uncertainty in linear control systems. The corresponding quest for theoretical and computational tools which tackle nonlinear systems has fallen short of expectations although gradual but steady progress has been made in several fundamental issues. Recently, a new framework called ``state-dependent scaling approach'' has been developed by this project conductor and it provides us with a new successful route to robustness issues in nonlinear systems control. An objective of this project is to tackle unsolved problems of nonlinear robust control by means of solutions to state-dependent scaling problems. A part of the project also includes the establishment of a more generalized theory encompassing broad classes of truly nonlinear and complex systems which have not been studied yet. Compared with theories of linear control, it is sometimes said that nonlinear control tends toward domain specific theories. It is reasonable in the sense that nonlinear control should tackle various nonlinearities causing diverse phenomena. This is a reason why analytic computation by hand or ad hoc symbolic computation by machines have dominated the practice of nonlinear control theories. We need a new theory which unifies the treatment of diverse nonlinearities. The new theory would provide us with a unified mathematical formulation which can be solved numerically in an efficient manner.

Theory of Sampled-Data Control

This project focuses on `non-synchronism' or `asynchronism' in sampled-data control systems. A sampled-data control system exhibits the non-synchronism when timing of control action and that of measurement acquisition are independent of each other. In practice, nonsynchronous implementation of sampled-data controllers often takes place unintentionally or intentionally. The non-synchronism may influence the control performance if the sampling frequency is slow relative to the plant or if the controller is decentralized. Recently, the use of vision sensor in feedback control has gained increased attention in mobile robot navigation, manipulation and autonomous driving. Hybrid sensor-based controllers which combine vision sensors with conventional sensors, such as potentiometer, rotary encoder and laser range sensor, are becoming a necessary strategy. The combination yields two levels possessing substantially different time constants. One includes intelligent processing of image information and calculation of the control inputs based on a bundle of visual information. The other includes the control based on conventional sensors. If one or few microprocessors are shared by the two levels, each level occupies the processors in turn. This time-sharing creates a phase shift among two control levels with different hold intervals, which results in a nonsynchronous system. Another common situation of non-synchronism also arises from large-scale interconnected systems such as communication network systems where distributed multiple processors are operated. We yet often encounter non-synchronism in process control where multirate control operates a huge plant consisting of systems having widely different time constants. Unfortunately the treatment of the non-synchronism is believed to be mathematically complex and delicate, and theoretical investigation of the non-synchronism has not received much attention, despite the existence of strong motivations in application. The fundamental goal of this project is to address those new issues and solve those problems mathematically.

Control of Magnetic Disk Drives

An important phenomenon which motivates this project is the enormous increase in capacity of computer storage devices, which has pushed its control performance into areas which until recently have been impossible. The control of magnetic disk drives is an important area of control where sophisticated mathematical tools of robust control theory is indispensable. Areal density of magnetic disk storages has been increasing by more than 60% every year. In other words, the track pitch has reached 100 nm recently. For the purpose of super swift and accurate head positioning, replacement of conventional single-actuator mechanism with dual-stage actuators has attracted much attention by engineers in advanced research and development sections. This project seeks an effective way of exploiting the potential of dual-stage actuators. The strategy of this project, which is believed to be an key to successful design, is cooperative multirate control. The most challenging point is how to circumvent the unique situation of magnetic disk drives which demand faster and more accurate positioning by feedback controllers with considerably less frequent acquisition of head position information.

Control of Wastewater Treatment Plants

Every modern society places great emphasis on environment conservation. There are increasing demands of technology which saves our earth. Control engineers are now trying to solve environmental problems arising from wastewater treatment. The progress is, however, still far from rapid development in other fields of engineering. This project is to develop a new technology of controlling biological wastewater treatment systems to make the effect of wastewater on the environment minimum in a cost effective way. In modern wastewater treatment plants, activated sludge processes have been used for the purpose of removing pollutants such as organic carbons, nutrients and some chemical compounds. In recent years, many governments are planing to introduce more comprehensive regulations including extensive removal of organic carbon, nitrogen and phosphorus. In order to meet such tight legislation, this project focuses on the configuration the Anaerobic-Anoxic-Oxic(A2O) process which places a numbers of biological reactors with and without aeration in an effective manner. The interconnected systems are highly nonlinear and complex in nature so that application of nonlinear feedback control and model reduction theories is inevitable in order to enhance the quality of wastewater treatment. The project is concerned with process analysis, modeling, simulation, diagnosis and control design.

Theory of Complex Large-Scale Systems

Given the current rate of technological innovation, the role of control in many engineering disciplines would increase dramatically in the near future. Most of future control applications will require sophisticated design tools which can accommodate large numbers of components in a system. Autonomy and coordination are crucial factors in controlling and designing complex large-scale systems. The project focuses on these issues theoretically. One of application-oriented themes is flow control of communication networks. Power control of wireless communication is also an important topic.

Analysis and Design of biological systems

This project is motivated by an analogy between artificial systems and biological systems, which has been attracted few bio-engineers until recently. This analogy is less understood especially in view of control mechanisms furnished automatically in every biological system. A main reason of this lack of progress seems to be that only few scientists were able to pay attention to dynamics. Importance of control theory for elucidating dynamics of biological processes has not been noticed properly yet. This project brings sophisticated theoretical and mathematical tools of control of nonlinear dynamics into bioscience, and they are exploited to do simulation, modeling and design of biological and metabolic systems. The analysis of sensitivity and robustness would is imperative in understanding functions of biological cell networks and extracting principles of biological systems. Theories of Nonlinear control and stability would play an important role in the research.

Control of Artful Machines

The project focuses on designing and building laboratory experimental machines which furnish unique or acrobatic movement, capability of various actions and control functions. An example is a double cart system which was developed in our laboratory several years ago. The double cart system has been an excellent test bed for demonstrating effectiveness of recently-developed theoretical tools of control, such as multirate sampled-data control, frequency separating cooperative control, robust nonlinear control. The project is challenging since experimental machines sometimes suffer from severe uncertainties and dynamic environments. This project is a great means for exposing students not only to usefulness of state-of-art control theories, but also to system engineering approaches to designing, building, managing, and maintaining complex systems.