First steps of transferring animal strides to a biomimetic soft walker

Abstract

The inherent benefits of soft materials in robotic designs have rendered soft robotics a growing field in research and engineering. Due to their compliance, soft robots are safe in working environments shared with humans, offer great potential in health care and medical applications, and may be operational in environments inaccessible or unfit for their solid body counterparts. However, for truly soft, self-contained robots, onboard electronics-free control is required. While there are pneumatic transistors which can be combined to simple control logics, the weight of these circuits may sometimes to overburden soft-legged robots. To overcome the weight limitation of our current soft robotic prototypes, we sought inspiration from nature by studying the leg morphology and parasagittal gait of mammals. They have been shaped by evolution to support the heaviest terrestrial animals on earth: elephants. We assume that the leg morphology and strides of elephants are optimized for energy efficiency and/or load bearing, and we translated their characteristics to a pneumatically actuated elephant soft robotic leg. However, as soft actuators are remarkably different from the mammal joint-and-muscle-system, a direct transfer from joint angles and muscle movement is not desirable. We therefore adapted the known kinematics of elephant strides to PneuNet bending actuators by means of approximating the actuators' bending angles to elephants' joint angles and subsequently arranging different actuator states into a sequence in order to approximate the elephant strides. We here present our current version of a biomimetic soft walker with parasagittal gait achieving a speed of 126 mm/s (0.82 body lengths per second) and a total load capacity of >5.2× its body weight.