Brain-inspired self-organizing modular structure to control human-like movements based on primitive motion identification

In this paper we will propose a modular structure to control the human-like movements of a robot in a way similar to which human brain does to perform the motor control. Modeling of human motor control and motor learning has attracted researchers׳ attention in robotics and artificial intelligence for decades. It is obvious that discovering the motor control functionality of brain as the most complex, sophisticated and powerful information-processing device leads to significant advancements in robots movement. Hence, our proposed modular controller is based on human brain behavior in using neural mechanisms named internal models and primitive motion identification which leads to extract and learn the latent simple motions in order to imitate observed complex movements. The study is accomplished based on formerly proposed structure, MOSAIC, which provides remarkable efficiency in motor control modeling. Examination of the proposed structure with real recorded data, confirms the performance of the controller in learning and executing motion tasks.