Predicting future object states using learned affordances

The notion of affordances that was proposed by J.J. Gibson, refers to the action possibilities offered to the organism by its environment. In a previous formalization, affordances are defined as general relations that pertain to the robot-environment interaction and they are represented as triples which consist of the initial percept of the environment, the behavior applied and the effect produced. In this paper, we focus on the object affordances and propose a developmental method that enables the robot to ground symbolic object-based operators in its own continuous sensory-motor experiences. The method allows the robot to learn the object affordance relations which can be used to predict the change in the percept of the object when a certain behavior is executed.