Closed-Loop Control and Perception Techniques for Soft Grippers

Pneumatic soft grippers have attracted considerable attention because their intrinsic compliance enables the safe manipulation of fragile and irregularly shaped objects. However, their practical performance remains constrained by challenges in structural design, proprioceptive and tactile perception, and closed-loop control. This review examines representative pneumatic actuation architectures, including PneuNet actuators, pneumatic artificial muscles, and fiber-reinforced actuators, and discusses how their structural and material characteristics govern deformation and grasping behavior. It further reviews flexible strain, pressure, optical tactile, and vision-based sensing approaches for real-time estimation of finger deformation, contact force, and incipient slip. The integration of these sensing modalities with adaptive and model-based control strategies is analyzed with respect to grasping accuracy, stability, and robustness. Finally, promising research directions are identified, including self-sensing materials, compact and energy-efficient actuators, multimodal perception, and data-driven control methods.