A Neural Network Feedforward PID Control Method for Soft Pneumatic Actuator Based on Origami Mechanism: Achieving Accurate Position Control

Soft pneumatic actuators offer better safety and flexibility compared to rigid actuators. However, they suffer from modeling difficulties and nonlinear delays, which prevent precise control. Research on the dynamic modeling and control of soft pneumatic actuators is a difficult subject. Neural networks are powerful tools for modeling complex dynamic systems. This essay suggests a PID control algorithm based on neural network feedforward compensation for precise position control of soft pneumatic actuators. By introducing a neural network to establish a feedforward dynamic compensation model, the algorithm predicts the linear displacement error of the soft pneumatic actuator in advance and compensates for it. This approach combines with a PID position controller to form a composite controller. The dynamic model of actuator is utilized to convert displacement into the amount of air required by the controller for precise control. The drive position experiments demonstrate that the control method is able to reach the drive expectations quickly and stably, thus demonstrating that the control method is able to effectively control soft pneumatic drives.