Human-Agent Team Based on Decision Matrices: Application to Road Traffic Management in Participatory Simulation

Abstract

AbstractThe main motivation for this work is the design of context-sensitive systems, involving a human-agent team. More precisely, the participants in the current context, are human beings and software agents, who have to coordinate themselves for the collaborative achievement of a complex task. We propose a model of interaction between these heterogeneous entities (humans, software agents) based on decision matrices inspired from game theory. The case study concerns the traffic management through intersections, involving several groups composed of human operators and software agents (playing the role of assistants), in a participatory simulation. An evaluation of the results obtained is presented in order to show the feasibility of the proposed approach. We show that despite its simplicity, the interaction model is relatively efficient, because it takes into account context characteristics and the potential availability of software agents.