Uncovering emotional memories in robot soccer players

Abstract

Memory is central to the emotional experience of playing sports. The capacity to recall great achievements, triumphs and defeats inevitably influences the emotional state of athletes and people in general. Nevertheless, research on robot competitions that has been striving to mimic real-world soccer, such as the well-known RoboCup challenge, never considered the relevance of memory and emotions, nor their possible connection. This paper proposes a data mining approach to emotional memory modelling with the purpose of replicating the link between emotion and memory in a Ro-boCup scenario. A model of emotional fluctuations is also proposed based on neurological disorders to investigate their effect on the robot's ability to choose appropriate behaviours. The proposed model is evaluated using the NAO robot on a simulation environment. By utilizing emotion to assess memories stored, NAO was able to successfully choose behaviours based on the optimal outcomes achieved in the past.