Efficient bipedal locomotion on rough terrain via compliant ankle actuation with energy regulation


Kerimoglu D., Karkoub M., Ismail U., MORGÜL Ö., SARANLI U.

BIOINSPIRATION & BIOMIMETICS, vol.16, no.5, 2021 (Journal Indexed in SCI) identifier identifier identifier

  • Publication Type: Article / Article
  • Volume: 16 Issue: 5
  • Publication Date: 2021
  • Doi Number: 10.1088/1748-3190/ac13b1
  • Title of Journal : BIOINSPIRATION & BIOMIMETICS
  • Keywords: bipedal locomotion, compliant ankle, passive dynamics, rough terrain, energy feedback, DYNAMIC WALKING, COMPASS-GAIT, OPTIMIZATION, STABILITY, DESIGN, MODEL, ENERGETICS, ROBOT

Abstract

Legged locomotion enables robotic platforms to traverse on rough terrain, which is quite challenging for other locomotion types, such as in wheeled and tracked systems. However, this benefit-moving robustly on rough terrain-comes with an inherent drawback due to the higher cost of transport in legged robots. The ultimate need for energy efficiency motivated the utilization of passive dynamics in legged locomotion. Nevertheless, a handicap in passive dynamic walking is the fragile basin of attraction that limits the locomotion capabilities of such systems. There have been various extensions to overcome such limitations by incorporating additional actuators and active control approaches at the expense of compromising the benefits of passivity. Here, we present a novel actuation and control framework, enabling efficient and sustained bipedal locomotion on significantly rough terrain. The proposed approach reinforces the passive dynamics by intermittent active feedback control within a bio-inspired compliant ankle actuation framework. Specifically, we use once-per-step energy regulation to adjust the spring precompression of the compliant ankle based on the liftoff instants-when the toe liftoffs from the ground-of the locomotion. Our results show that the proposed approach achieves highly efficient (with a cost of transport of 0.086) sustained locomotion on rough terrain, withstanding height variations up to 15% of the leg length. We provide theoretical and numerical analysis to demonstrate the performance of our approach, including systematic comparisons with the recent and state-of-the-art techniques in the literature.