The Second Joint International Seminar on Applied Analysis and Synthesis of Complex Systems
|
|
| IIASA - Technische Universität München - Kyoto University The Second Joint International Seminar on Applied Analysis and Synthesis of Complex Systems |
| June 30-July 1, 2005 |
| Technische Universität München, München, Germany |
Human is a complex mechanical system and the ultimate goal of robotics is to realize some of these complex functions of humans by engineering manner. We focus on "manipulation", which is a kind of physical interactions between the hand and an object being manipulated, as one of the complex functions that humans have. It is very important to understand fundamental property of human functions before trying to build a robot. We have been developing teleoperation systems and haptic virtual reality systems not only to apply them to some useful domains but also to use them as a research platform to investigate human manipulation functions and understand what the bottleneck to realize these functions is. In this talk, we will introduce some resent teleoperation systems and haptic interfaces that we developed.
Keywords: Interaction, Control, Human-Maschine System, Metamodeling, Situation Operator Model, Cognitive Systems, Adaption, Automated Supervision
The contribution deals with the development and application of a suitable language / modeling approach to describe the Human-Machine-Interaction. After a short introduction into the Situation-Operator Modeling approach as methodological background, the contribution details the distinctions between algorithms (and human control) on one hands side and classical control on the other. Using the developed approach a strong relation between control and information science appears from a structural point of view. The classification also shows that between both, a lot of other realizations (or from the point of view of the flexible human 'control': restrictions) exist, whereby the well known classical technical approach appears as the simpliest one (or from the point of view of the flexible human 'control': the version with the most restrictions). The items of the classifications will be defined and detailed in the contribution with the examples of PID-control, Optimal control, Algorithms, Human Interaction, and Intelligent Systems. This gives - beside the academic scheme - the view to the next steps of improving automatic control algorithms. Based on microcontrollers, databases and intelligent data pre-processing some 'human control'-qualities like learning and flexible response abilities as well as situated control could be imitated by technical realizations including cognitive approaches. The examples span a wide area for new types of technical realizations to Human and Machine Supervision and to the Realization of Autonomous Systems.
Turbulence is a complex state of fluid motion. The flow field varies randomly both in space and in time, and never repeats. This makes it very difficult to extract any universal dynamical properties of turbulence from the instantaneous flows. There is no way to pick up, with confidence, any representative parts of turbulent flows from a finite series of temporal evolution. Thus, it would be nice if there are some reproducible flows, or skeletons of turbulence, which represent the turbulent state well. This is reminiscent of unstable periodic orbits in chaotic dynamical systems. The chaotic attractor contains infinitely many unstable periodic orbits. Some statistical properties associated with a strange attractor are described in terms of the periodic orbits embedded in it. Since unstable to small perturbations, such periodic orbits are not found by forward integration but can be captured by Newton-Raphson iterations, as we have succeeded recently. Here, we present two examples of periodic motions, each of which represents the properties of the respective turbulent state well by itself. One example is given for a plane Couette system, which represents the typical generation cycle of turbulent activity, i.e. the repetition of alternate generation and breakdown of streamwise vortices and low-speed streaks. The other is for isotropic turbulence, which reproduces the Kolmogorov energy spectrum in the universal range.
Fluid flows have a complex structure that incorporates both regions of apparent random behaviour, and coherent, long-lived vortices. This mixture means that fluid flows have complex transport properties, in terms of advection of heat and other active and passive scalars. In particular coherent vortices can trap scalars, and smooth out fluctuations on a range of time-scales, controlled by the interaction of molecular diffusion and of the reduction of scales by stretching and folding in the flow. The talk will give an overview of these transport properties.
The derivation of fluid-dynamic systems from the Boltzmann system by means of a systematic asymptotic analysis for small Knudsen numbers (or small mean free paths) is reviewed. An example of such fluid-dynamic systems, composed of the compressible Euler equations and numerically constructed boundary conditions, that describes steady flows of a vapor around its condensed phases in the presence of a small amount of a noncondensable gas is presented, and it is shown that, in the fluid-dynamic limit (the limit in which the Knudsen number vanishes), the noncondensable gas with an infinitesimal average concentration has a significant effect on the overall vapor flows. The system is then applied to some concrete problems, which demonstrate the above-mentioned anomalous effect explicitly.
The problem of deriving the classical equations of hydrodynamics (such as the Euler and Navier-Stokes equations) from the kinetic theory of gases goes back to Maxwell's papers, and was later formalized by Hilbert. This talk will review recent progress from the mathematical viewpoint on this classical problem.
With increasing demand for the complexity and accuracy of the 3D microstructure shape for MEMS, process controllability of three dimensional (3D) micro fabrication technologies such as Deep RIE of single crystal silicon (Si), LIGA and anisotropic etching of Si have been becoming more and more crucial. One of the promising approaches to address this requirement is the exploitation of a computational MEMS process design and process development. The computational analysis such as FEM of the static and dynamic behavior of the MEMS has been already widely utilized owing to the effort of a software development which can handle the objects with multi-physics at multi-scale. In contrast, there is few example for the effort of computational MEMS process design and process development due to the several reasons such as immaturity of the MEMS processes and lack of understanding about governing process physics. The effort to exploit the computational MEMS process design and development is less than necessary for the future development of MEMS. As an example of the computational MEMS process design and development under study, a deep X-ray lithography process design and development for 3D microstructures fabrication will be presented.
To date, coupled simulation enviroments for MEMS devices have been implemented in commerical and free software. However, the structural optimization as applied to multiphysics is not well discussed to date. In most cases, the optimization algorithm merely follows the traditional single-field problem case, in which the design variables and sensitivity are updated sequencially. This paper presents a promising methodology to implement structural topology optimization via a fully coupled partial differential equation (PDE) expression. The multiphysical structural optimization is integrated through derivation of specified Lagrangian-Euler equations. A benchmark example of structural topology optimization which contains coupled effects is presented in order to demostrate the feasibility of this methodology.
Unique nano-morphology and properties of obliquely deposited thin films have been extensively investigated since the end of the 1950s. Recently, thin films with highly controlled isolated columns such as helix and zigzag have been developed. Various shape-related properties such as optical and magnetic anisotropies and mechanical compliance have been observed. In this talk, the growth mechanism and morphology design of the columnar thin films will be reviewed. In addition, our recent attempts for the optical and mechanical applications of these films will be discussed.