Measuring Morphological Computation in Turtle-Inspired Multi-Modal Robotic Locomotion

Biological systems exploit the dynamic interaction between their body morphology and the environment to adapt more effectively to uncertain and changing environments. Amphibious creatures such as turtles are particularly impressive in their ability to seamlessly traverse radically different environments, e.g., water to sand. There is a growing interest in better understanding how the morphology can help to facilitate this interaction with the goal to formalize the underlying principles and translate them into better robot designs. This approach is often referred to as Morphological Computation (MC) and recently theoretical approaches have been introduced to quantify the contribution of the morphology during locomotion and interaction with the environment. However, these theoretical models have primarily been used in simulations rather than in physical models that interact with complex real-world environments. In this study, we apply the quantification of MC to a physical robotic flipper inspired by a marine turtle to experimentally evaluate the contribution of morphology during both crawling and swimming gaits in sand and water, respectively. The results show a clear difference in average MC magnitudes, clearly indicating the changing influence of the flipper morphology during locomotion in these two environments. MC was found to be lower while in contact with the sand, indicating that a larger contribution through active control was required to counteract friction. This study highlights how the relative contributions of robot morphology and active control can be experimentally determined for improved multi-modal locomotion.