Sensorimotor Prediction with Neural Networks on Continuous Spaces

In the context of Developmental Robotics, we propose to learn how the sensations of a robot are modified by its action. Many theories of Artificial Intelligence argue that sensorimotor prediction is a fundamental building block of cognition. In this paper, we learn the sensorimotor prediction on data captured by a mobile robot equipped with distance sensors. We show that Neural Networks can learn the sensorimotor regularities and perform sensorimotor prediction on continuous sensor and motor spaces.