Application of robot autobiographical memory in long-term human-robot social interactions

Abstract

Robots are becoming a part of the human society. They are intended to maintain long-term social interactions with humans. A memory about each interaction partner is essential for a robot for maintaining long-term social interactions. Therefore, robot memory must possess the abilities for extracting, storing, updating and recalling information during social interactions. This work studies the application of a robot’s autobiographical memory in long-term social interactions in multi-user environments. The proposed system is capable of acquiring knowledge about users through friendly conversations and recall them during future interactions. The autobiographical memory which has been designed with a three-layered architecture is employed for storing significant information which is required for long-term social interactions. The interaction manager module has been designed as a finite state machine for managing the interaction between the robot and the user. Dialogue flows were defined for each state to communicate with the user. The system has been tested and validated by using a set of experiments.