Spine morphology and energetics: how principles from nature apply to robotics

Abstract

Inspired by the locomotive advantages that an articulated spine enables in quadrupedal animals, we explore and quantify the energetic effect that an articulated spine has in legged robots. We compare two model instances of a conceptual planar quadruped: one with a traditional rigid main body and one with an articulated main body with an actuated spinal joint. Both models feature four distinct legs, series elastic actuation, distributed mass in all body segments, and limits on actuator torque and speed. Using optimal control to find the energetically optimal joint trajectories, actuator inputs, and footfall timing, we examine and compare the positive mechanical work cost of transport of both models across multiple gaits and speeds. Our results show that an articulated spine increases the maximum possible speed and improves the locomotor economy at higher velocities, especially for asymmetrical gaits. The driving factors for these improvements are the same mechanistic effects that facilitate asymmetrical gaits in nature: improved leg recirculation, elastic energy storage in the spine, and enlarged stride lengths.