Time-Aware Long-term Episodic Memory for Recurring HRI.

In this paper, we present a time-aware entity-based episodic memory for recurring Human-Robot Interaction tasks. Entities are asynchronously updated, while an importance factor modulates forgetting, in order to facilitate search and management of the stored information. Time awareness and forgetting are crucial for long-term HRI, since they facilitate robot behavior adaptation based on past experiences. The performance of the proposed memory schema is experimentally evaluated within a recurring breakfast serving scenario.