Catálogo de publicaciones - libros

Recent developments in logistics show an increasing trend towards autonomous control. Intelligent software agents, that represent logistic objects like shipping containers, plan and schedule their way through a logistic network. This paper addresses the aspect of cooperation. A special focus lies on the second step of the model for cooperation, namely team formation. The question is by which criteria shipping containers, or logistic objects and agents in general, can form clusters. Starting from the particular demands of the logistics domain this paper argues in favour of conceptual, spatial, and temporal properties. A framework that takes concept, location, and time into account is introduced and demonstrated by an example application.

In real-life multi-agent planning problems, long-term plans will often be invalidated by changes in the environment during or after the planning process. When this happens, short-term operational planning and scheduling methods have to be applied in order to deal with these changed situations. In addition to the dynamic environment, in such planning systems we also have to be aware of sometimes conflicting interests of different parties, which render a centralized approach undesirable. In this paper we investigate two agent-based scheduling architectures where stakeholders are modelled as autonomous agents. We discuss this approach in the context of an interesting airport planning problem: the planning and scheduling of deicing and anti-icing activities. To coordinate the competition between agents over scarce resources, we have developed two mechanisms: one mechanism based on , and one based on a more traditional (Vickrey) auction. Experiments show that the auction-based mechanism best respects the preferences of the individual agents, whereas the decommitment mechanism ensures a fairer distribution of delay over the agents.

Pedestrian simulation is a challenging and fruitful application area for agent-based modeling and simulation in the traffic and transportation domain. In this paper we will present the concepts and results of a particular project study: an agent-based simulation of pedestrian traffic of the complete railway station of Bern during the most busy morning hours. Overall more than 40 000 agents are passing through during 1,5 virtual hours. Going beyond traditional approaches for microscopic pedestrian simulation, our simulated pedestrians are not only capable of moving without collisions between two pre-defined locations, but are able to flexibly plan and re-plan their way through the railway station. A short glance and some discussion about the potential of agent-based pedestrian simulation closes this contribution.

Failures in plan execution can be attributed to errors in the execution of plan steps or violations of the plan structure. The structure of a plan prescribes which actions have to be performed and which precedence constraints between them have to be respected. Especially in multi-agent environments violations of plan structure might easily occur as the consequence of synchronization errors. While in previous work we have concentrated on the first type of failures, in this paper we introduce the idea of diagnosing plan structure violations. Using a formal framework for plan diagnosis, we describe how Model-Based Diagnosis can applied to identify these violations of plan structure specifications and we analyze their computational complexity.

The paper addresses the problem of recovering the execution of a multi-agent plan when the occurrence of unexpected events (e.g. faults) may cause the failure of some actions. In our scenario actions are executed concurrently by a group of agents organized in teams and each agent performs a local control loop on the progress of the sub-plan it is responsible for. When an agent detects an action failure, the agent itself tries to repair (if possible) its own sub-plan and if this local recovery fails, a more powerful recovery strategy at is invoked. Such a strategy is based on the cooperation of agents within the same team: the agent in trouble asks another teammate, properly selected, to cooperate for recovering from a particular action failure. The cooperation is aimed at achieving the goal assigned to the agents’ team despite the action failure and to this end the agents exchange sub-goals and synthesize new plans.

We study here the behaviour of a trust and reputation model for agents in P2P environment, where agents act as providers or consumers of resources. We present partial results of the refinement process of our model in order to identify the suitable structure, functions, and parameters to correctly manage trust and reputation. We show some experimental results comparing several alternatives to obtain the quality of the response given the requirements, and the increase of the satisfaction of the recommended response when it uses a right way to estimate trust and reputation values using stored information about similar tasks.

Since the last decade reputation management has been examined as a possible foundation for trust establishment and trust dissemination in distributed artificial intelligence. While the systems are distributed the reputation management systems are in most cases centralized. This paper describes the implementation of the ReGreT approach to trust and reputation management in Multi-Agent Systems in a decentralized environment, where the services which provide reputation management are also distributed.