Self-Organized Map Based Learning System for Estimating the Specific Task by Simple Instructions
This paper discusses a learning system for service robots to estimate specific tasks by using simple instructions from human beings. Intelligent robots are expected to operate in human living areas, so service robots should understand specific tasks from simple instructions given by human beings. It is important to perceive environmental situations and to adapt to human preferences. We propose a learningmethod using the Self-Organized Map (SOM) to estimate specific tasks from both human behavior measurement but also environmental measurement. Through simulation experiments, we verified that the proposed SOMbased method considers environmental situations associated time variations and show that service robots decide table-clearing tasks according to human intent.