Central Pattern Generator Control of a Tensegrity Swimmer

Abstract

Rhythmic motion employed in animal locomotion is ultimately controlled by neuronal circuits known as central pattern generators (CPGs). It appears that these controllers produce efficient, oscillatory command signals by entraining to an efficient or economic gait via sensory feedback. This property is of great interest in the control of autonomous vehicles. The objective of this study is to experimentally validate synthesized CPG control of a tensegrity swimmer. The prestressed cables in a tensegrity structure provide a method of simultaneous actuation and sensing, analogous to the biological motor control mechanism of regulating muscle stiffness through motoneuron activation and sensing the resulting motion by stretch receptors. A three cell, class 2 tensegrity swimmer is designed and built, and open-loop control tests characterize its swimming performance. We then determine gaits for desired entrainment, and use a graphical design method to construct and test the closed-loop system. Lastly, we perform perturbed tests of the swimmer to illustrate the robustness of CPG control.