Lenko Grigorov

Near-Optimal Online Control of Dynamic Discrete-Event Systems

A class of time-varying discrete-event systems, named dynamic discrete-event systems, is defined. The goal of this paper is to provide a method which is modular and can be applied in real solutions for the optimization of the online control of such systems. First, a simple control algorithm is presented, followed by illustrative examples of different issues that can arise if it is used. Afterward, a more complicated near-optimal online control algorithm with normalization of string values is proposed. The time variability of the systems is accounted for and the average computational time is drastically reduced. This is demonstrated with a set of simulations of the performance of the new algorithm.

Establishing remote conversations through eye contact with physical awareness proxies

We present a mechanism for initiating mediated conversations through eye contact. An eyePHONE is a physical proxy of a remote individual that senses and conveys attention using an eye tracking device and a pair of actuated eyeballs. Users may initiate calls by jointly looking at each others eyePHONE. We discuss how this allows participants to implement some of the basic social rules of face-to-face conversations in mediated conversations.

Conceptual Design of Discrete-Event Systems Using Templates

This work describes the research conducted in the quest for designing better software for discrete-event system (DES) control. The think-aloud data from an exploratory observational study of solving DES control problems, together with other relevant research, led to the proposal of a novel approach to DES problem solving called the template design methodology. This methodology does not require the introduction of new control theory; it is rather an reinterpretation of the existing modelling framework. It provides a high-level overview of a DES design, and facilitates the use of template models. Software supporting this methodology was implemented as an extension to existing DES software. The methodology and the implementation were subsequently evaluated using 12 subjects. Significant improvements in the speed of problem solving as well as positive evaluations by the subjects were observed. The usability data do not show any drawbacks to applying the methodology.

Design of discrete-event systems using templates

A new methodology for the design of DES control is proposed which allows for the creation of high-level conceptual designs by using encapsulated low-level elements. The approach can be used within the standard framework of modular supervisory control. The notion of DES templates is introduced, where typical behaviors for both DES modules and specifications are represented in an abstract way. The control engineer creates instances of these abstractions and specifies the way the instances interact. System modeling and the design of specifications occur simultaneously. Speed and robustness of the design process are improved since there is no need to consider details or to reimplement similar parts of a system. The proposed methodology is applied to a small robotic testbed to get real-world feedback.

Problem solving in control of discrete-event systems

Human performance in problem solving in the context of supervisory control of discrete-event systems is discussed. Software packages for such problems usually offer only rudimentary support in the form of visual data entry and algorithmic computations. Other activities such as system modeling and verification of the solution have to be done without computer support. This work is a first step in understanding how humans approach discrete-event control problems. An observational study is described, where subjects are video-taped solving supervisory control problems. One video recording is used to construct a taxonomy of the task and then the different activities of the subject are encoded accordingly. The data are analysed to discover patterns of human activity. These results can be used to guide the design of software for computer-aided problem solving in this context.

ISSUES IN OPTIMAL CONTROL OF DYNAMIC DISCRETE-EVENT SYSTEMS

Abstract We define the notion of Dynamic Discrete-Event Systems, a class of time-varying systems, and present a simple approach to optimal control of such systems. More specifically, we use limited-lookahead online control and an algorithm which tries to maximize the benefit of the executed sequences of events, while at the same time ensuring that unwanted (illegal) sequences are avoided. We use examples to illustrate the different types of problems that can arise if such control is used, for example, overspecialization and failure to take advantage of available resources. These issues are used to formulate a list of desirable properties for a new algorithm that optimizes the control of dynamic systems.

Techniques for the Parametrization of Discrete-Event System Templates

Abstract Two techniques for the parametrization of templates in the template design methodology are introduced. First, the use of finite-state automata extended with variables is demonstrated. This kind of parametrization allows one to vary the occurrences of events by varying a parameter. Second, a new way of parametrizing templates is proposed, named compositional parametrization. With this technique, a template consists of the composition of a number of structurally identical components whose events are indexed differently, depending on a parameter. This kind of parametrization allows one to vary the structure and event set of the template by varying a parameter. An enhancement of compositional parametrization with the use of special selector transitions is also discussed. This extension allows one to model interactions between components which depend on the total number of components in a composition and thus cannot be modelled by simple parametrization of event indices. With the choice of a specific parameter, regular finite-state automaton models are obtainable from the templates parametrized with any of the aforementioned techniques. Thus, such parametrized templates can be easily introduced in the template design modelling environment and they integrate seamlessly with the supervisory control framework. Motivating examples to illustrate the application of template parametrization are provided.

Template design and automatic generation of controllers for industrial robots

The basic theory of supervisory control of discrete-event systems is extended with the notion of templates, which simplifies the modeling of controllers since one can work with conceptual designs. In this work, software which provides support for the new design approach is presented along with its application to a robotic testbed.

Dynamic discrete-event systems with instances for the modelling of emergency response protocols

A type of model, dynamic discrete-event systems with instances, is proposed for use with time-varying systems where a number of system components may share common behavior. The motivation for this model comes from the area of emergency response protocols where systems may change unexpectedly, however, there are predefined types of actors. The model is based on previous work on dynamic discrete event systems and on template design. Synchronization patterns are introduced which allow the generic definition of component interactions regardless of a particular system composition. The synchronous product operation is updated in order to employ these patterns. Advantages of the proposed model include compactness of representation, preservation of the identity (history) of individual components and, with the use of online control, amenability to dynamic changes of control specifications.