Decentralized Control for Morphology-Adaptive Gait Generation in Sprawling Quadruped Locomotion

Quadruped animals orchestrate their bodily degrees of freedom to achieve agile and adaptive locomotion. Sprawling locomotion, characterized by walking gaits with lateral trunk bending, presents an intriguing model behavior for investigating underlying motor control principles. This locomotion pattern is observed across amphibians, reptiles, and mammals. Moreover, paleontological evidence suggests that this behavior was also present in extinct primitive tetrapods, indicating that sprawling locomotion likely encapsulates fundamental principles for whole-body coordination in quadruped motor control. A key characteristic of sprawling locomotion is its adaptive gait patterns. For instance, salamanders modulate both their footfall patterns and axial movements according to speed. They employ a walking gait at slow speeds and transition to a walking trot gait at higher speeds. Similarly, their axial flexion manifests as a standing wave at slower speeds and transforms into a traveling wave at higher speeds. Additionally, axial movements adapt in response to morphological features. Species with short trunks exhibit traveling axial motion with long wavelengths relative to their body length, whereas species with long trunks display motion patterns with shorter wavelengths. These adaptive gait generations represent a fundamental property of motor control in sprawling locomotion. This study aims to investigate the underlying control mechanisms for adaptive motor control in sprawling locomotion through mathematical modeling and computational simulations. In our previous work, we proposed a decentralized control framework utilizing sensory feedback for axial-limb coordination, successfully reproducing speed-dependent gait transitions in a simulated model. Based on this work, we developed a control mechanism capable of morphology-adaptive gait generation and considered shared principles for whole-body coordination in quadruped motor control.