Agentes polimórficos dinâmicos inteligentes: teoria e aplicação

Abstract

When multiple agents with different characteristics are considered to solve a set of problems, it is very challenging to specify the number of agents, the specific function of each one to be used, and the schedule of these actions in order to solve these problems. Different situations may demand different sets of agents with specific knowledge regarding how to solve each problem. To deal with scenarios like this, the present article suggests an innovation at the Intelligent Agent Theory, a new concept called Intelligent Dynamic Polymorphic Agent (IDPA). The IDPA has the goal of trying to find the minimum set of resources and agents (each one with its own characteristics) to plan and execute specific problems using AI planning. By on-line actualization, the IDPA architecture allows the quest for available agents present in the system and being feed by information about their respective domains. Using this knowledge, the system uses a predefined heuristic model to dynamically assist the planner obtaining a plan with a reduced number of resources. It also identifies and dispatches the necessary agents in order to carry out the plan. After this identification, the agents merge with the entity called APDI, which, through polymorphism, is capable of accomplish the initial task Another benefit of this methodology is that, due to its dynamic behavior, it is possible to change the original plan by the advent of an unexpected problem, even if the plan is being already executed.