Catálogo de publicaciones - libros

Characterised by geographical, temporal, functional and/or semantic distribution, today’s enterprise models engage multiple design teams with heterogeneous skills cooperating together in order to achieve global optima in design. The success of this distributed design organization depends on critical factors such as the efficient management of the design related information circulated in the distributed environment and the support for the necessary cooperation process among participants dispersed across the enterprise. This paper proposes a multi-agent design information management system to support the synthesis and presentation of information to distributed teams for the purposes of enhancing design, learning, creativity, communication and productivity. Autonomous software agents and information ontologies enable the proposed system facilitating interoperation among distributed resources as well as knowledge sharing, reuse and integration.

A persistent publish/subscribe messaging model allows the creation of an application-independent fault-tolerant layer for multi-agent systems. We propose a layer which is capable of supporting heterogenous agent platforms from different vendors. This layer is a three-tier application, which is accessible from multi-agent systems via web-services or a persistent publish/subscribe messaging system. We describe the design of the fault-tolerant layer, its messaging system, as well as the algorithm of fault-recovery procedure in the case of agent and/or host death. We also present performance analysis of the proposed solution, to justify its use in systems which demand different levels of reliability.

In this paper, we present SWATS which supports task-oriented automated teleoperation of Web-based devices. The proposed system employs Semantic Web Services technology and AI planning technique to achieve operational automaticity.

Today’s system administrator is forced to perform individual configuration and maintenance tasks (i.e. ) on numerous systems, in various formats. These tasks are time-consuming and labor intensive. Several research projects have attempted to resolve these issues with the development of an integrated, centralized management system. However, many tasks are still left to the system administrator for manual handling. A customized configuration system that reflects comprehensive context has not yet been fully realized. This paper proposes a context aware self-configuration system by employing multi-agents to collectively gather contextual information based on the system resources and user’s system usage patterns. This proposed system, then analyzes the collected information and performs automatic configuration as and when required. This system will allow not only enable automation of the previously manual tasks, but also allow, in effect, a more customized configuration.

Agent Technology has become very popular in the last few years as a new approach to developing software systems. Multi Agent Systems (MAS), a term used to describe the incorporation of multiple types of agents into various systems, is a way of designing and implementing a system with the advantages of agent entities. We chose to use agents as a decision support tool for use in a Retail Inventory Management System. Since the management of inventory is crucial to the success of most companies, and since we see a potential major role for agents in the business process management MAS seems a likely choice for a decision support platform. This work stems from our prior work in simulating a MAS inventory system, then implementing the system for production use.

This paper presents a framework for planning and scheduling integration based on a uniform constraint-based representation. Such representation is inspired to time-line based planning but has the unique characteristic of conceiving both resource and causal constraints as abstract specifications that generate segments of temporal evolution to be scheduled on the time-line. This paper describes the general idea behind this type of problem solving, shows how it has been implemented in a software architecture called , and presents an example of application for the generation of mission planning commands for automating the management of spacecraft operations.

This paper deals with plan execution on agents/robots in highly dynamic environments. Besides a formal semantics of plan execution and a representation of plans as programs, we introduce the concept of plan invariants. Plan invariants are similar to loop invariants in imperative programs in that they have to be true during the whole plan execution cycle. Once a plan invariant fails the plan execution is stopped and other plans that are more appropriate in the current context are considered for execution instead. The use of plan invariants allows for an early detection of problems. Plan assumptions that are required for a plan to succeed are explicitly represented by plan invariants.

When using a constructive search algorithm, solutions to scheduling problems such as the job shop and open shop scheduling problems are typically represented as permutations of the operations to be scheduled. The combination of this representation and the use of a constructive algorithm introduces a bias typically favouring good solutions. When ant colony optimisation is applied to these problems, a number of alternative are available, each of which interacts with this underlying bias in different ways. This paper explores both the structural aspects of the problem that introduce this underlying bias and the ways two pheromone representations may either lead towards poorer or better solutions over time. Thus it is a synthesis of a number of recent studies in this area that deal with each of these aspects independently.

The paper proposes an implementation of the population learning algorithm (PLA) for solving the permutation flowshop scheduling problem (PFSP). The PLA can be considered as a useful framework for constructing a hybrid approaches. In the proposed implementation the PLA scheme is used to integrate evolutionary, tabu search and simulated annealing algorithms. The approach has been evaluated experimentally. Experiment has produced 14 new upper bounds for the standard benchmark dataset containing 120 PFSP instances and has shown that the approach is competitive to other algorithms.

In order to realize a navigation system for refugees in disaster areas, we must reduce computation costs required in setting escape routes. Thus, in this paper, we propose a method for reducing the costs by grasping whole danger regions in a disaster area from a global perspective. At first, we estimate future changes of dangerous regions by a simple way and link all regions with Danger Levels. Then, we index estimated dangerous regions by extended R-tree. In this step, we link the Danger Levels with depths of the extended R-tree and each Danger Level is managed at each depth of the extended R-tree. Finally, we show how our approach effects in setting escape routes.