Modeling of Hybrid Reasoning of Dynamic Agent in ABAsim Architecture

The paper proposes a mechanism to change reasoning model of ABAsim architecture's dynamic agent. The design process of agent's reasoner component is discussed. After a review of different approaches to model agent's reasoning and goal assignment, a mechanism to register multiple reasoners and to change the reasoner while respecting modeling principles of ABAsim architecture (Agent Based Architecture of simulation models) is presented. By expanding the set of available reasoners, the dynamic agent gains the capability to select the most suitable reasoning engine and thus adapt to different situations and needs, e.g. higher performance or need to adapt or learn in a dynamic environment. The proposed approach has been implemented in the simulation tool PedSim developed at the Faculty of Management Science and Informatics, University of Žilina. Various scenarios have been created to demonstrate performance impact and functionality of the proposed mechanism.