Design and Control of a Soft-Robotic Tentacle Using Shape Memory Alloys

Abstract

This work presents the design and control of a soft-robotic tentacle actuated by shape memory alloy (SMA) springs, inspired by the muscular arrangement of an octopus arm. A desktop-scale prototype consisting of three antagonistically actuated sections was developed, providing six rotational degrees of freedom. To achieve proportional control of the SMA actuators, a baseline PID controller was compared with a model-based feedforward–feedback (FF–FB) controller incorporating an inverse Preisach hysteresis model. The Preisach model was identified and experimentally validated on a dedicated singleactuator test bench. Results show that the model accurately captures major hysteresis loops but exhibits significant errors during minor loops and cannot account for rate-dependent hysteresis behavior. Consequently, the FF–FB controller underperformed relative to the PID controller, which demonstrated robust and reliable tracking on both the test bench and the complete tentacle prototype. These findings confirm that proportional closed-loop control of an SMA-actuated multi-section tentacle is feasible, while highlighting that high control accuracy remains challenging due to the non-linear, hysteretic nature of SMA actuators. Overall, SMAs offer a compact, silent, and mechanically simple actuation solution, with bandwidth and control accuracy identified as the primary limitations for future development.