Embodying rather than encoding: Towards developing a source-filter theory for undulation gait generation

Abstract

Biological undulation enables legless creatures to move naturally, and robustly in various environments. Consequently, many kinds of undulating robots have been developed. However, the fundamental mechanism of biological undulation gait generation has not yet been well explained, which hinders deepening the investigation and optimization of these robots. Towards developing a theory for explaining this biological behavior, which will further guide the design of artificial undulation systems, we propose a hypothesis based on both biological findings and previous robotics studies. To verify the hypothesis, we investigate embodied intelligence of undulation locomotion via a mechanical system. Through experimental study, we observe the phenomenon that undulation gait is a production of the source, which is the torque inputs, and the filter, which is the natural dynamics of the system. We further derive a general mathematical model and conduct morphological computation accordingly. From a simple model to a complicated system, our work explores the principles of undulation gait generation. Our findings significantly simplify the control system design of artificial undulating systems.