Nowadays scientists are concerned about the topic, to determine how the Central Nervous System (CNS) controls the muscles in order to move the limbs into a desired target position. N. Bernstein showed that the system has numerous degrees of freedom (DoF) and the system is highly redundant, which makes it a really difficult problem. However, it was proven that every movement can achieved by finite set of elementary motor components, called primitives.
The major goal of this thesis is to transfer principles of human motor control to humanoid robots using motion primitives. Humans and robots are quite different in their sensing capabilities, their mechanical structure, and the way they process data. This thesis focuses on the reaching movement of the arm, which point is to achieve a human-like arm motion with a humanoid robot, and also investigates some leg motions. For this purpose I create a movement database with the help of the Computer and Automation Research Institutes (SZTAKI) of Hungarian Academy and Sciences (MTA) 3D cave, the Virtual Collaboration Arena (VirCA) virtual environment, and a data suit. Next I define the motion primitives using principal component analysis, and try to recreate the original movement using these primitives.