A sensorless soft pneumatically powered actuator

Soft actuators have gained considerable attention in recent years because of their inherent flexibility and adaptability. Presently, the majority of these actuators require control laws for closed-loop bending control that encompass both angular displacement and angular velocity. Nevertheless, incorporating curvature sensors into flexible actuators poses several obstacles. Estimating the rotational velocity in flexible systems presents substantial challenges. This study proposes a method for addressing the above issue by introducing a strategy for designing, manufacturing, modeling, and controlling soft pneumatically powered actuators, all without the need for a curvature sensor. The elimination of a curvature sensor dramatically simplifies the production process of flexible mechanisms and substantially expands its possibilities for use. This article uses an experimental approach to develop a second-order polynomial that maps air pressure to bending angle using the Euler-Bernoulli principle. Afterward, the motion dynamics model is integrated with an air pressure controller to achieve a specified bending angle. During operation, the soft actuator experiences variations in the volume of the air chamber, which in turn affects the compressed air pressure. Nevertheless, precisely measuring this fluctuation is a substantial difficulty. The Fuzzy Extended State Observer (FESO) and the Sliding Mode Controller (SMC) collaborate to effectively control and maintain the air pressure at the required level. The extended state observer (ESO) is able to accurately predict external disturbances and model inaccuracies. In using fuzzy logic with the ESO, it is crucial to achieve a compromise between the accuracy of the estimation and the reduction of noise. Subsequently, a sliding mode controller is employed to eliminate the expected disturbance in its entirety. The comparison of simulation and experimental results provides evidence of the effectiveness of this control method in tracking the bending angle trajectory. The controller shows a good response when the steady-state error is within a range of less than 10°.