A soft perceptive multimodal locomotion robot via double-helix fiber-reinforced artificial muscles

Abstract

Pneumatic artificial muscle has a high payload-to-weight ratio and rapid response with rich form factors and great design flexibility, making it a promising candidate for muscle-driven bioinspired robots. Here, we propose a perceptive pneumatic artificial muscle (PPAM) with self-length sensing via a seamlessly integrated double-helix conductive fiber coil. The double-helix coil fiber restricts the radial expansion of the silicone tube to achieve axial extension in actuation and measures its own length with a resolution of 10 μm by monitoring its self-inductance. A perceptive multimodal locomotion robot using three PPAMs as a bundle was developed, with crawling, turning, lateral movement, and rolling capabilities. By applying machine learning to analyze the three-channel inductance changes caused by body deformation under different loading conditions, the PPAM-based soft robot can recognize different external stimuli with an average accuracy of 99.28%. It is demonstrated that the soft robot can recognize a deep pit and bypass it automatically, change its gait amplitude to crawl through a low-roof tunnel, and sense its body posture change when a side flip occurs and automatically rolls back to its original position. In summary, the robot can perceive its own state and environment like a biological worm and can actively adjust its locomotion gait/mode to navigate itself through diverse terrains.