Measuring multimodal deformations in soft inflatable actuators using embedded strain sensors

Abstract

The intrinsic mechanical compliance that makes soft robotic systems ideal for safe, adaptive physical interactions in the presence of uncertainty also poses significant challenges in control. The highly-nonlinear deformations that soft robots undergo are difficult to model and predict, and are even harder to measure and modulate when the soft structures are exposed to complex mechanical loads. This paper presents a method for tracking the deformation modes of soft inflatable bending actuators used in grasp assist devices. Soft strain sensors are designed and strategically placed within the body of a bending actuator to allow measurement of the primary deflection mode - flexion/extension - and measurement of axial twist and lateral deflection modes which are induced by grasping forces. Multimodal bending sensor prototypes were tested independently and embedded in pneumatic bending actuators. Results demonstrated varying levels of measurement accuracy and highlighted the challenges of tracking motion in soft devices. Multimodal sensors were also evaluated in a virtual proprioception experiment and demonstrated efficacy in providing sensory capabilities for human augmentation devices.