A new biped locomotion involving arms swing based on neural network with harmony search optimizer

This paper deals with a novel model free approach composed of coupled adaptive nonlinear and linear oscillators with emphasis on make robot's walking more stable and faster. In this regard Matsuoka neural oscillators have been used to generate control signals to control the locomotion of a humanoid robot. To reach smoother walking and increase speed and robustness, the current system controls roll of arms during locomotion. To find the best angular trajectory and optimize neural oscillator parameters, a new population-based search algorithm, called the Harmony Search (HS) algorithm, has been used. We implement the method on the simulated NAO robot that has been implemented in Robocup 3D soccer simulation environment. Simulation Results show that using from hands during walking can make walking fast and robustness and also, harmony search algorithm is fast in convergence.