“…Given a higher-level controller (e.g., admittance force control) yielding desired trajectories (e.g., as seen in other soft wearable assistive devices [26]), a low-level controller operating at the pneumatic-system level needs to ensure the desired trajectory is achieved. In our past work we have developed feedforward [17], [27] as well as setpoint [28] and learning-based feedback [29] controllers for soft pneumatic actuators used in various iterations of our wearable device prototypes [17], [23]. Feedback controllers can help make the device more responsive, while tracking desired setpoints is a key function on its own, e.g., to counter gravity at the shoulder joint [3].…”