Walking is like slithering: A unifying, data-driven view of locomotion

Area of Science:

• Robotics and Biomechanics
• Locomotion Science
• Geometric Mechanics

Background:

• Legged locomotion is common in nature and robotics, but modeling slipping contacts is challenging.
• Existing simulation capacities are limited for detailed multi-contact slipping models.
• Understanding diverse locomotion methods is key for advanced robotics.

Purpose of the Study:

• To present a unifying principle for multilegged walking, slithering, and low Reynolds number swimming.
• To develop data-driven kinematic models for locomotion analysis.
• To explore the fundamental mechanics governing legged movement, including slipping.

Main Methods:

• Generated data-driven, principally kinematic models of locomotion.
• Analyzed low-slip animal (Argentine ant) and high-slip robotic systems (hexapod, multi-leg robots).
• Investigated the relationship between body shape, velocity, and turning rate.

Main Results:

• Principally kinematic models explained variability in body velocity and turning rate.
• Models successfully predicted walking behaviors outside training data.
• Walking mechanics were found to be principally kinematic regardless of leg number, slipping, or turning rate.

Conclusions:

• Grounded walking, with or without slipping, follows principally kinematic equations of motion.
• A unified model for swimming, slithering, and walking is achievable through geometric mechanics.
• Findings offer insights into the evolution of locomotion control and new approaches for robotic motion planning.