Enriching Contact Information Through Fiber Bragg Gratings-Based Exteroception in Soft Bending Actuators

Tactile sensing is of utmost importance for accomplishing fine manipulation tasks and extracting crucial physical features in soft grippers. A contact force estimation and localization approach based on sensing modalities using Fiber Bragg Grating (FBG) transducers and artificial intelligence models are proposed for pneumatically actuated soft fingers. The small packaged sensor unit enables modular design, ease of assembly, and high repeatability with negligible effects on the mechanical performance of the soft finger. The proposed neural networks process the output of the sensing modality and mitigate errors from material nonlinearity, fabrication, and assembly of soft fingers. This, in turn, enhances the accuracy and transferability of force and position estimations, leading to proper metrological indicators. These results provide a step forward in the development of smart soft grippers enabling them to attain “transparent” exteroception. As a result, this work contributes to empowering soft grippers with skillful capabilities to acquire precise tactile information for safe manipulation ranging from agricultural to biomedical fields.