An extended BDI model for human behaviors: decision-making, learning, interactions, and applications

Modeling human decision-making behaviors under a complex and uncertain environment is quite challenging. The goal of this tutorial is to discuss an extended Belief-Desire-Intention (BDI) framework that the authors' research group has been developing last decade to meet such a challenge, integrating models and techniques (e.g. Bayesian Belief Network, Decision Field Theory, Depth First Search) available in the fields of engineering, psychology, computational science, and statistics. First, major modules of the extended BDI framework are discussed, where those modules represent cognitive functions (i.e. perception, goal seeking, planning, decision-making, execution) of an individual. Then, two extensions are considered, where the first one involves dynamic evolution of underlying modules over time (i.e. learning and forgetting), and the second one involves human interactions (e.g. competition, collaboration, compromise, accommodation, avoidance). To illustrate the proposed approach, various applications are used, such as emergency evacuation during bomb attack, driver and pedestrian behaviors, and cyber social network.