Dual-SLIP model based galloping gait control for quadruped robot: A Task-space Formulation

Abstract

This paper presents a novel locomotion control framework that achieves stable galloping gait for a torque-controlled quadruped robot. By analytically exploiting the stance dynamics of the Spring-Loaded Inverted Pendulum (SLIP) model, a two-layered Dual-SLIP model based Task-space Formulation (DS-TSF) is developed to control the 12-DoF quadruped robot with an active spine. On the higher layer, a dead-beat controller based on the derived Approximate Apex Return Map (AARM) with guaranteed high prediction accuracy is devised to provide desired apex state. This reference SLIP-like behavior serves as the target Center of Mass (CoM) trajectories of the fore- and hind-body of the quadruped robot. On the lower layer, a prioritized multi-task controller is further developed to enforce the dual-CoMs of the fore- and hind-body to behave following the target dynamics of two uncoupled SLIP hoppers on sagittal plane. The compatible motion control of the active spinal joint is fulfilled on the null-space of the prior task without generating confliction. The simulation results have demonstrated the effectiveness of the proposed locomotion control method.