A model for an affective non-expensive utility-based decision agent

Abstract

We present an affective model for an autonomous decision agent implementable within non-expensive robotic platforms. The agent makes decisions influenced by affective factors when interacting with humans and other agents under the Adversarial Risk Analysis framework. The model is based on a multi-objective additive utility function, with weights partly regulated by the agent's emotional states and mood. Our motivation for this work is not to provide a more accurate descriptive model, but rather to use affection as a basic element within a decision-making process that could lead to improved performance of a robotic agent, possibly making it more attractive. This would facilitate its adoption in application areas such as companion pets to provide emotional support to the elderly or kids with autism, or as educational support within an edutainment framework.