A Self-Organizing Network with Varying Density Structure for Characterizing Sensorimotor Transformations in Robotic Systems

In this work, we present the development of a neuro-inspired approach for\ncharacterizing sensorimotor relations in robotic systems. The proposed method\nhas self-organizing and associative properties that enable it to autonomously\nobtain these relations without any prior knowledge of either the motor (e.g.\nmechanical structure) or perceptual (e.g. sensor calibration) models.\nSelf-organizing topographic properties are used to build both sensory and motor\nmaps, then the associative properties rule the stability and accuracy of the\nemerging connections between these maps. Compared to previous works, our method\nintroduces a new varying density self-organizing map (VDSOM) that controls the\nconcentration of nodes in regions with large transformation errors without\naffecting much the computational time. A distortion metric is measured to\nachieve a self-tuning sensorimotor model that adapts to changes in either motor\nor sensory models. The obtained sensorimotor maps prove to have less error than\nconventional self-organizing methods and potential for further development.\n