Sensory-driven neck-limb coordination mechanisms for walk-trot-gallop gait transitions

Quadruped animals exhibit various gait patterns and switch from one to another depending on their speed. These gait patterns involve complex whole-body coordination and are mainly controlled by a distributed neural network consisting of central pattern generators (CPGs) and peripheral sensory feedback. However, how these neural networks generate various types of whole-body coordination remains unclear. Here, we focus on horse locomotion and present a neural network model that reproduces walk-trot-gallop transitions by changing only a single parameter related to speed. This model uses ground reaction force (GRF) and trunk angular velocity as sensory information. The simulation results demonstrated that the proposed feedback mechanisms provide self-organized neck-limb coordination in response to gait patterns. Gait evaluation based on speed, neck swing amplitude, peak GRF, and gait cycle stability shows similar tendencies to those of actual animals. These findings suggest that our model could be valuable for understanding adaptive whole-body coordination mechanisms and advancing robot control design.