Efficiency on legged locomotion pattern using adaptive frequency Hopf oscillator

Abstract

This paper presents results on an adaptive controller coupled with a stretchable pendulum including a hybrid dynamic model in order to find the resonance frequency given by the dynamics of the mechanism. This idea has been originated from biological neuro-musculo-skeletal system that is abundantly exploited in bio-inspired robotic prototypes. The spring-mass model was incorporated with pendulum model that both ones are extremely referred as templates for legged locomotion. Interaction with the ground caused that the model involves with two dynamic phases that are switched to each other through transition events. This hybrid model in coupling with adaptive frequency Hopf oscillator in feedback loop was simulated and control parameters was adjusted via genetic algorithm to enhance the performance of the achieved motion pattern. The obtained results reveal that the controller is able to excite the mechanical system in an energy efficient pattern by way of exploiting resonance phenomenon. Also, adaptation of system to environment changes is studied and seen that the controller is able to find the resonance state after changes.