Synthesis of Neural Oscillator based Dynamic Rhythmic Generation in Quadruped Robot Locomotion

Abstract

This paper proposes motion pattern generation based sensory motor coordination in quadruped robot. Quadruped animal generates their motion pattern dynamically in every level of speed. The proposed locomotion model implements combined sensory motor coordination system as the dynamic walking pattern generation and polynomial equation as the swinging model in Cartesian level. In the sensory motor coordination model, motor neural oscillator as the rhythmic signal generator will be interconnected with internal sensory neurons, which composed by ground touch sensors in the sole of feet and body pose movement sensor (Inertial Measurement Unit). In order to optimized the interconnection structure between sensory neurons and motor neurons, we used multi-objective evolutionary algorithm. Error of desired walking and energy required is minimized as fitness function. In order to prove the effectiveness of the model, we implemented it in computer simulation and in proposed Felidae-like quadruped robot. The proposed model succeeded generate dynamic motion pattern in different movement speed. The proposed model has similarity motion pattern with animal gait in certain movement speed. The movement changing behavior depending on the changes of desired velocity are also shown.