Closed-Loop Control of Fiber-Braided Soft Actuator via Digital Flex Sensing and GA-Optimized PID

Soft pneumatic actuators facilitate safe and adaptable interaction, but their nonlinear and hysteretic characteristics present notable difficulties in achieving accurate modeling and real-time control. This study proposes a closed-loop control framework for a Fiber Braided Bending Actuator (FBBA) incorporating a digital capacitive flex sensor (Bend Labs One-Axis) for angular feedback. A third-order Auto-Regressive with Exogenous Input (ARX) model was developed using a black-box system identification approach in MATLAB Simulink, yielding a model fit of 92.36% against experimental validation data. A Proportional-Integral-Derivative (PID) controller tuned using a Genetic Algorithm (GA) was implemented to regulate actuator position. Simulation results demonstrated improved transient performance over manual tuning, with reduced rise and settling times and zero overshoot. A qualitative comparison with a conventional analog resistive sensor (Spectra Symbol FS-L-0095), under identical control conditions, showed that the digital sensor enabled smoother tracking and reduced signal fluctuation. Real-time experiments using step and multi-step inputs were conducted, demonstrating high control accuracy and fast response. These findings support the effectiveness of combining digital sensing, data-driven modeling, and evolutionary control techniques for soft robotic systems that demand precise deformation tracking.