CPG-based locomotion control of a quadruped robot with an active spine

Central pattern generators (CPGs) are neural networks re- sponsible for producing rhythmic behaviours and are commonly found in both vertebrate and invertebrate animals. This paper proposes a novel internal feedback mechanism for a CPG model designed to generate leg- spine coordinated locomotion in a quadruped robot with an active spine. This mechanism enables the CPG to independently control the frequency and amplitude of the stance and swing durations while also modifying the definition of stance and swing phases to generate more gaits. The CPG model’s results are demonstrated on a simulated ”tensegrity quadruped robot” called TQbot, which features a flexible spine with 3 degrees of freedom (DOF). By adjusting the parameters, the CPG model can gen- erate gaits with leg-spine coordination and uses the spine for turning.