Training AIBO like a dog - preliminary results

Abstract

This work describes a method to facilitate human adaptation to a pet robot. A pet robot learns which behavior it could execute when stimulus are given and a human user learns how to give commands to the robot through its various sensors. A pet robot utilizes a computational classical conditioning model for learning to interpret human's commands. We propose and implement such a mutual adaptation framework, and develop like-dog heuristics to facilitate the human's adaptation to a pet robot AIBO. Finally we evaluate the heuristics through preliminary experiments.