Design of a Trainable Controller Inspired from Neural System to Generate Complex Behaviors in Humanoid Robots

In this paper we introduced a new method for motion control in humanoid robots. The problem of movement learning specially dance and repetitive actions of human beings to humanoid robots is a major challenge in the field of robotics. Imitation learning, which is a subset of supervised learning, is a main form to teach complex tasks to the humanoid robot, and the accordingly is based on that an artificial system can imitate a lot of information through learning from human trainer. The main technique is using Central Pattern Generators structures which is able to produce required motion trajectories based on imitation learning. Systematic design of this these neural networks is main problem which is solved in this paper. The proposed model is a basic paradigm for imitation learning in the humanoid robots which do not required direct design of controller and programming. The proposed model has many benefits including smooth walking patterns and modulation during imitation. Simulation results of this learning system in the robot simulator (WEBOTS) that has been linked with MATLAB software and its implementation on a NAO robot demonstrate that the robot has learned desired motion with high accuracy. This model can be extended and used in the Nao soccer player both for the standard platform and the 3D soccer simulation leagues of Robocup SPL competitions to train different types of motions.