Visual Patterns Recognition in Robotic Platforms Through the Use of Neural Networks and Image Processing

Current design trends with respect to the manufacture of human-like robotic platforms are constantly growing; day to day manufacturers generate new platforms in order to improve the integration of these technological tools into our daily activities. This paper proposes a methodology associated with pattern recognition by using Adaptive Resonance Theory (ART) Neural Networks and image processing techniques, allowing this way the NAO robotic platform to be easily used in experiments of human-robot interactions. It attempts to integrate an algorithm of geometric figures recognition so the robotic NAO can identify shape, color and size for the patterns in a fast and efficient way, to be associated with established activities or actions, such as: controlled displacements, extremities movement or played voice messages.