Modulation of Central Pattern Generators (CPG) for the Locomotion Planning of an Articulated Robot

Abstract

The present paper proposes the approach of locomotion in mammals to be applied in articulated robotics. This is achieved using Central Pattern Generators by amplitude modulation of oscillatory signals to communicate the angle of rotation of each of the joints that are involved in a specific type of locomotion. Performing simulations to determine viability by frequency and amplitude, a better response was found in the amplitude modulation. A series of locomotion data with dogs were compiled and used as a reference for the amplitude modulation of the differential equation systems that replicate the Central Pattern Generators of the articulations of the quadrupedal robot. Recurrent neural networks in continuous time were used to represent the CPG. The angle was modulated as a function of the amplitude of the cyclic signal produced by the Central Pattern Generators allowing to manage the setpoints (angles) for a given articulation, between 0° and 90°. Other works, although related to Central Pattern Generators, and some focused on reproducing the model, none of them deals with the construction of modulated signals that represent joint angles based on data obtained from biomechanical studies of locomotion by quadrupeds. A distributed autonomous control architecture based on modular and hierarchical Central Pattern Generators, organized in two layers, that simultaneously synchronizes and executes the movement of each joint from each leg, and for the total movement production is proposed.