Actuation mechanisms and machine learning-based optimization for the twisted and coiled artificial muscles: a review

Abstract The rapid advancement of artificial intelligence has spurred widespread interest in artificial muscles as core actuating components for soft robotics. Twisted and coiled artificial muscles (TCAMs), a novel class of soft actuators fabricated by twisting and coiling high-performance fibers, exhibit notable advantages, including low cost, high flexibility, long cycle life, large actuation strain, and high energy density. The TCAMs can generate multiple motion modes, including bending, twisting, contraction/extension, and buckling in response to external stimuli. The TCAMs demonstrate significant potential for application in fields such as soft robotics and wearable assistive devices. This review comprehensively examines the developmental trajectory and current research landscape of TCAMs, synthesizes progress in actuation strategies, actuation mechanisms, and machine learning applications, and concludes with perspectives on future research directions. We further assess persistent challenges in hysteresis suppression and manufacturing reproducibility, proposing cross-disciplinary solutions toward deployable artificial muscle systems.