AFFORDANCE MODEL-BASED DISCOVERY OF SUBGOALS AND TASK LEARNING, AND SYSTEM THEREOF

摘要

PURPOSE: An affordance model-based discovery of intermediate goal and task learning, and a system thereof are provided to implement speedy learning and reuse an intermediate goal by obtaining information that a robot is chanted to manage an object or non-object and generating it as the intermediate goal. CONSTITUTION: In an affordance model-based discovery of intermediate goal and task learning, and a system thereof, a visual converter(10) extracts a feature vector related to a task of a robot from learning data through the teachings or the simulation. An appetitive state transition model generating unit(25) produces a task study results based on the feature vector through an appetitive state transition model. The appetitive state transition model generates an action cause the appetitive state so that an object or non-object induces an action.